Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research Estimating global injuries morbidity and mortality: methods and data used in the Global Burden of Disease 2017 study ►►Additional material is Spencer L James,1 Chris D Castle,1 Zachary V Dingels,1 Jack T Fox,1 Erin B Hamilton,1 published online only. To view, 1 1 1 1 please visit the journal online Zichen Liu, Nicholas L S Roberts, Dillon O Sylte, Gregory J Bertolacci, (http://dx.​ ​doi.org/​ ​10.1136/​ ​ 1 1 1 2 injuryprev-2019-​ ​043531). Matthew Cunningham, Nathaniel J Henry, Kate E LeGrand, Ahmed Abdelalim, 3 4 5 For numbered affiliations see Ibrahim Abdollahpour, Rizwan Suliankatchi Abdulkader, Aidin Abedi, end of article. Kedir Hussein Abegaz,6,7 Akine Eshete Abosetugn,8 Abdelrahman I Abushouk,9 10 11 12,13 Correspondence to Oladimeji M Adebayo, Jose C Adsuar, Shailesh M Advani, Marcela Agudelo-­Botero,14 Tauseef Ahmad,15,16 Muktar Beshir Ahmed,17 Received 14 October 2019 18,19 20 21 Revised 29 November 2019 Rushdia Ahmed, Miloud Taki Eddine Aichour, Fares Alahdab, Accepted 6 December 2019 22 23 24,25 Published Online First Fahad Mashhour Alanezi, Niguse Meles Alema, Biresaw Wassihun Alemu, 24 August 2020 Suliman A Alghnam,26 Beriwan Abdulqadir Ali,27 Saqib Ali,28 Cyrus Alinia,29 Vahid Alipour,30,31 Syed Mohamed Aljunid,32,33 Amir Almasi-Hashiani­ ,34 Nihad A Almasri,35 Khalid Altirkawi,36 Yasser Sami Abdeldayem Amer,37,38 Catalina Liliana Andrei,39 Alireza Ansari-Moghaddam,­ 40 Carl Abelardo T Antonio,41,42 Davood Anvari,43,44 Seth Christopher Yaw Appiah,45,46 Jalal Arabloo,30 Morteza Arab-­Zozani,47 Zohreh Arefi,48 Olatunde Aremu,49 Filippo Ariani,50 Amit Arora,51,52 Malke Asaad,53 Beatriz Paulina Ayala Quintanilla,54,55 Getinet Ayano,56 Martin Amogre Ayanore,57 Ghasem Azarian,58 Alaa Badawi,59,60 Ashish D Badiye,61 Atif Amin Baig,62,63 Mohan Bairwa,64,65 Ahad Bakhtiari,66 Arun Balachandran,67,68 Maciej Banach,69,70 Srikanta K Banerjee,71 Palash Chandra Banik,72 Amrit Banstola,73 Suzanne Lyn Barker-­Collo,74 Till Winfried Bärnighausen,75,76 Akbar Barzegar,77 http://injuryprevention.bmj.com/ Mohsen Bayati,78 Shahrzad Bazargan-­Hejazi,79,80 Neeraj Bedi,81,82 Masoud Behzadifar,83 Habte Belete,84 Derrick A Bennett,85 Isabela M Bensenor,86 Kidanemaryam Berhe,87 Akshaya Srikanth Bhagavathula,88,89 Pankaj Bhardwaj,90,91 Anusha Ganapati Bhat,92 Krittika Bhattacharyya,93,94 Zulfiqar A Bhutta,95,96 Sadia Bibi,97 Ali Bijani,98 Archith Boloor,99 Guilherme Borges,100 Rohan Borschmann,101,102 Antonio Maria Borzì,103 Soufiane Boufous,104 Dejana Braithwaite,105 Nikolay Ivanovich Briko,106 Traolach Brugha,107 108 109,110 111 112 Shyam S Budhathoki, Josip Car, Rosario Cárdenas, Félix Carvalho, on September 27, 2021 by guest. Protected copyright. João Mauricio Castaldelli-Maia,­ 113 Carlos A Castañeda-­Orjuela,114,115 Giulio Castelpietra,116,117 Ferrán Catalá-López,118,119 Ester Cerin,120,121 Joht S Chandan,122 Jens Robert Chapman,123 Vijay Kumar Chattu,124 Soosanna Kumary Chattu,125 Irini Chatziralli,126,127 Neha Chaudhary,128,129 Daniel Youngwhan Cho,130 Jee-­Young J Choi,131 Mohiuddin Ahsanul Kabir Chowdhury,132,133 Devasahayam J Christopher,134 Dinh-Toi­ Chu,135 Flavia M Cicuttini,136 João M Coelho,137 Vera M Costa,112 Saad M A Dahlawi,138 Ahmad Daryani,139 Claudio Alberto Dávila-­Cervantes,140 Diego De Leo,141 Feleke Mekonnen Demeke,142 Gebre Teklemariam Demoz,143,144 © Author(s) (or their 23 145,146 147 employer(s)) 2020. Re-­use Desalegn Getnet Demsie, Kebede Deribe, Rupak Desai, permitted under CC BY. 148 149 150 Published by BMJ. Mostafa Dianati Nasab, Diana Dias da Silva, Zahra Sadat Dibaji Forooshani, 151 152 153 154 To cite: James SL, Castle CD, Hoa Thi Do, Kerrie E Doyle, Tim Robert Driscoll, Eleonora Dubljanin, 155,156 157,158 159 Dingels ZV, et al. Inj Prev Bereket Duko Adema, Arielle Wilder Eagan, Demelash Abewa Elemineh, 2020;26:i125–i153.

James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 i125 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research Shaimaa I El-­Jaafary,2 Ziad El-­Khatib,160,161 Christian Lycke Ellingsen,162,163 Maysaa El Sayed Zaki,164 Sharareh Eskandarieh,165 Oghenowede Eyawo,166,167 Pawan Sirwan Faris,168,169 Andre Faro,170 Farshad Farzadfar,171 Seyed-­Mohammad Fereshtehnejad,172,173 Eduarda Fernandes,174 Pietro Ferrara,175 Florian Fischer,176 Morenike Oluwatoyin Folayan,177 Artem Alekseevich Fomenkov,178 Masoud Foroutan,179 Joel Msafiri Francis,180 Richard Charles Franklin,181,182 Takeshi Fukumoto,183,184 Biniyam Sahiledengle Geberemariyam,185 Hadush Gebremariam,87 Ketema Bizuwork Gebremedhin,186 Leake G Gebremeskel,143,187 Gebreamlak Gebremedhn Gebremeskel,188,189 Berhe Gebremichael,190 Getnet Azeze Gedefaw,191,192 Birhanu Geta,193 Agegnehu Bante Getenet,194 Mansour Ghafourifard,195 Farhad Ghamari,196 Reza Ghanei Gheshlagh,197 Asadollah Gholamian,198,199 Syed Amir Gilani,200,201 Tiffany K Gill,202 Amir Hossein Goudarzian,203 Alessandra C Goulart,204,205 Ayman Grada,206 Michal Grivna,207 Rafael Alves Guimarães,208 Yuming Guo,136,209 Gaurav Gupta,210 Juanita A Haagsma,211 Brian James Hall,212 Randah R Hamadeh,213 Samer Hamidi,214 Demelash Woldeyohannes Handiso,185 Josep Maria Haro,215,216 Amir Hasanzadeh,217,218 Shoaib Hassan,219 Soheil Hassanipour,220,221 Hadi Hassankhani,222,223 Hamid Yimam Hassen,224,225 Rasmus Havmoeller,226 Delia Hendrie,56 Fatemeh Heydarpour,227 Martha Híjar,228,229 Hung Chak Ho,230 Chi Linh Hoang,231 Michael K Hole,232 Ramesh Holla,233 Naznin Hossain,234,235 Mehdi Hosseinzadeh,236,237 Sorin Hostiuc,238,239 Guoqing Hu,240 Segun Emmanuel Ibitoye,241 Olayinka Stephen Ilesanmi,242 Leeberk Raja Inbaraj,243 Seyed Sina Naghibi Irvani,244 M Mofizul Islam,245 Sheikh Mohammed Shariful Islam,246,247 Rebecca Q Ivers,248 Mohammad Ali Jahani,249 Mihajlo Jakovljevic,250 Farzad Jalilian,251 Sudha Jayaraman,252 Achala Upendra Jayatilleke,253,254 Ravi Prakash Jha,255 Yetunde O John-­Akinola,256 Jost B Jonas,257,258 Kelly M Jones,259 Nitin Joseph,260 Farahnaz Joukar,220 Jacek Jerzy Jozwiak,261 Suresh Banayya Jungari,262 Mikk Jürisson,263 Ali Kabir,264 Amaha Kahsay,87 Leila R Kalankesh,265 Rohollah Kalhor,266,267 Teshome Abegaz Kamil,268 Tanuj Kanchan,269 Neeti Kapoor,61 Manoochehr Karami,270 Amir Kasaeian,271,272 Hagazi Gebremedhin Kassaye,23 Taras Kavetskyy,273,274 Gbenga A Kayode,275,276 Peter Njenga Keiyoro,277 278 279 280 281 Abraham Getachew Kelbore, Yousef Saleh Khader, Morteza Abdullatif Khafaie, Nauman Khalid, http://injuryprevention.bmj.com/ Ibrahim A Khalil,282 Rovshan Khalilov,283 Maseer Khan,284 Ejaz Ahmad Khan,285 Junaid Khan,286 Tripti Khanna,287,288 Salman Khazaei,270 Habibolah Khazaie,289 Roba Khundkar,290 Daniel N Kiirithio,291 Young-Eun­ Kim,292 Yun Jin Kim,293 Daniel Kim,294 Sezer Kisa,295 Adnan Kisa,296 Hamidreza Komaki,297,298 Shivakumar K M Kondlahalli,299 Ali Koolivand,300 Vladimir Andreevich Korshunov,106 Ai Koyanagi,301,302 Moritz U G Kraemer,303,304 Kewal Krishan,305 Barthelemy Kuate Defo,306,307 Burcu Kucuk Bicer,308,309 Nuworza Kugbey,310,311 Nithin Kumar,312 Manasi Kumar,313,314 Vivek Kumar,315 Narinder Kumar,316 Girikumar Kumaresh,317 Faris Hasan Lami,318 Van C Lansingh,319,320 Savita Lasrado,321 Arman Latifi,322 323 324 325 326,327 136

Paolo Lauriola, Carlo La Vecchia, Janet L Leasher, Shaun Wen Huey Lee, Shanshan Li, on September 27, 2021 by guest. Protected copyright. Xuefeng Liu,328 Alan D Lopez,1,102,329 Paulo A Lotufo,330 Ronan A Lyons,331 Daiane Borges Machado,332,333 Mohammed Madadin,334 Muhammed Magdy Abd El Razek,335 Narayan Bahadur Mahotra,336 Marek Majdan,337 Azeem Majeed,338 Venkatesh Maled,339,340 Deborah Carvalho Malta,341 Navid Manafi,342,343 Amir Manafi,344 Ana-Laura­ Manda,345 Narayana Manjunatha,346 Fariborz Mansour-­Ghanaei,220 Mohammad Ali Mansournia,347 Joemer C Maravilla,348 Amanda J Mason-­Jones,349 Seyedeh Zahra Masoumi,350 Benjamin Ballard Massenburg,130 Pallab K Maulik,351,352 Man Mohan Mehndiratta,353,354 Zeleke Aschalew Melketsedik,194 Peter T N Memiah,355 Walter Mendoza,356 Ritesh G Menezes,357 Melkamu Merid Mengesha,358 Tuomo J Meretoja,359,360 Atte Meretoja,361,362 Hayimro Edemealem Merie,363 Tomislav Mestrovic,364,365 Bartosz Miazgowski,366,367 Tomasz Miazgowski,368 Ted R Miller,56,369 G K Mini,370,371 Andreea Mirica,372,373 Erkin M Mirrakhimov,374,375 Mehdi Mirzaei-­Alavijeh,251 Prasanna Mithra,260 Babak Moazen,376,377 Masoud Moghadaszadeh,378,379 Efat Mohamadi,380 Yousef Mohammad,381 Aso Mohammad Darwesh,382 Abdollah Mohammadian-­Hafshejani,383 Reza Mohammadpourhodki,384 Shafiu Mohammed,75,385 Jemal Abdu Mohammed,386 Farnam Mohebi,171,387 Mohammad A Mohseni Bandpei,388 Mariam Molokhia,389 i126 James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research Lorenzo Monasta,390 Yoshan Moodley,391 Masoud Moradi,392,393 Ghobad Moradi,394,395 Maziar Moradi-­Lakeh,396 Rahmatollah Moradzadeh,34 Lidia Morawska,397 Ilais Moreno Velásquez,398 Shane Douglas Morrison,130 Tilahun Belete Mossie,399 Atalay Goshu Muluneh,400 Kamarul Imran Musa,401 Ghulam Mustafa,402,403 Mehdi Naderi,404 Ahamarshan Jayaraman Nagarajan,405,406 Gurudatta Naik,407 Mukhammad David Naimzada,408,409 Farid Najafi,410 Vinay Nangia,411 Bruno Ramos Nascimento,412 Morteza Naserbakht,413,414 Vinod Nayak,415 Javad Nazari,416,417 Duduzile Edith Ndwandwe,418 Ionut Negoi,419,420 Josephine W Ngunjiri,421 Trang Huyen Nguyen,231 Cuong Tat Nguyen,422 Diep Ngoc Nguyen,423,424 Huong Lan Thi Nguyen,422 Rajan Nikbakhsh,425,426 Dina Nur Anggraini Ningrum,427,428 Chukwudi A Nnaji,418,429 Richard Ofori-­Asenso,430,431 Felix Akpojene Ogbo,432 Onome Bright Oghenetega,433 In-­Hwan Oh,434 Andrew T Olagunju,435,436 Tinuke O Olagunju,437 Ahmed Omar Bali,438 Obinna E Onwujekwe,439 Heather M Orpana,440,441 Erika Ota,442 Nikita Otstavnov,408,443 Stanislav S Otstavnov,408,444 Mahesh P A,445 Jagadish Rao Padubidri,446 Smita Pakhale,447 Keyvan Pakshir,448 Songhomitra Panda-­Jonas,449 Eun-­Kee Park,450 Sangram Kishor Patel,451,452 Ashish Pathak,453,454 Sanghamitra Pati,455 Kebreab Paulos,456 Amy E Peden,182,457 Veincent Christian Filipino Pepito,458 Jeevan Pereira,459 Michael R Phillips,460,461 Roman V Polibin,462 Suzanne Polinder,211 Farshad Pourmalek,463 Akram Pourshams,464 Hossein Poustchi,464 Swayam Prakash,465 Dimas Ria Angga Pribadi,466 Parul Puri,286 Zahiruddin Quazi Syed,91 Navid Rabiee,467 Mohammad Rabiee,468 Amir Radfar,469,470 Anwar Rafay,471 Ata Rafiee,472 Alireza Rafiei,473,474 Fakher Rahim,475,476 Siavash Rahimi,477 Muhammad Aziz Rahman,478,479 Ali Rajabpour-Sanati,­ 480 Fatemeh Rajati,392 Ivo Rakovac,481 Sowmya J Rao,482 Vahid Rashedi,483 Prateek Rastogi,446 Priya Rathi,233 Salman Rawaf,338,484 Lal Rawal,485 Reza Rawassizadeh,486 Vishnu Renjith,487 Serge Resnikoff,488,489 Aziz Rezapour,30 Ana Isabel Ribeiro,490 Jennifer Rickard,491,492 Carlos Miguel Rios González,493,494 Leonardo Roever,495 Luca Ronfani,390 Gholamreza Roshandel,464,496 Basema Saddik,497 Hamid Safarpour,498 Mahdi Safdarian,499,500 S Mohammad Sajadi,501 Payman Salamati,500 Marwa R Rashad Salem,502 Hosni Salem,503 Inbal Salz,504 505 506,507 508,509 510,511 Abdallah M Samy, Juan Sanabria, Lidia Sanchez Riera, Milena M Santric Milicevic, http://injuryprevention.bmj.com/ Abdur Razzaque Sarker,512 Arash Sarveazad,513 Brijesh Sathian,514,515 Monika Sawhney,516 Mehdi Sayyah,517 David C Schwebel,518 Soraya Seedat,519 Subramanian Senthilkumaran,520 Seyedmojtaba Seyedmousavi,521 Feng Sha,522 Faramarz Shaahmadi,523 Saeed Shahabi,524 Masood Ali Shaikh,525 Mehran Shams-­Beyranvand,526 Aziz Sheikh,527,528 Mika Shigematsu,529 Jae Il Shin,530,531 Rahman Shiri,532 Soraya Siabani,533,534 Inga Dora Sigfusdottir,535,536 Jasvinder A Singh,537,538 Pankaj Kumar Singh,539 Dhirendra Narain Sinha,540,541 Amin Soheili,542,543 Joan B Soriano,544,545 Muluken Bekele Sorrie,546 Ireneous N Soyiri,547,548 Mark A Stokes,549 Mu’awiyyah Babale Sufiyan,550 Bryan L Sykes,551 552,553 554 555 556,557

Rafael Tabarés-­Seisdedos, Karen M Tabb, Biruk Wogayehu Taddele, Yonatal Mesfin Tefera, on September 27, 2021 by guest. Protected copyright. Arash Tehrani-­Banihashemi,396,558 Gebretsadkan Hintsa Tekulu,559 Ayenew Kassie Tesema Tesema,560 Berhe Etsay Tesfay,561 Rekha Thapar,312 Mariya Vladimirovna Titova,178,562 Kenean Getaneh Tlaye,563 Hamid Reza Tohidinik,347,564 Roman Topor-Madry,­ 565,566 Khanh Bao Tran,567,568 Bach Xuan Tran,569 Jaya Prasad Tripathy,90 Alexander C Tsai,570,571 Aristidis Tsatsakis,572 Lorainne Tudor Car,573 Irfan Ullah,574,575 Saif Ullah,97 Bhaskaran Unnikrishnan,260 Era Upadhyay,576 Olalekan A Uthman,577 Pascual R Valdez,578,579 Tommi Juhani Vasankari,580 Yousef Veisani,581 Narayanaswamy Venketasubramanian,582,583 Francesco S Violante,584,585 Vasily Vlassov,586 Yasir Waheed,587 Yuan-P­ ang Wang,113 Taweewat Wiangkham,588 Haileab Fekadu Wolde,400 Dawit Habte Woldeyes,589 Temesgen Gebeyehu Wondmeneh,386 Adam Belay Wondmieneh,186,590 Ai-Min­ Wu,591 Grant M A Wyper,592 Rajaram Yadav,286 Ali Yadollahpour,593 Yuichiro Yano,594 Sanni Yaya,595 Vahid Yazdi-F­ eyzabadi,596,597 Pengpeng Ye,598 Paul Yip,599,600 Engida Yisma,601 Naohiro Yonemoto,602 Seok-Jun­ Yoon,292 Yoosik Youm,603 Mustafa Z Younis,604,605 Zabihollah Yousefi,606,607 Chuanhua Yu,608,609 Yong Yu,610 Telma Zahirian Moghadam,30,611 Zoubida Zaidi,612 Sojib Bin Zaman,132,613 Mohammad Zamani,614 Hamed Zandian,611,615 Fatemeh Zarei,616 Zhi-­Jiang Zhang,617 Yunquan Zhang,618,619

James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 i127 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research Arash Ziapour,533 Sanjay Zodpey,620 Rakhi Dandona,1,329,621 Samath Dhamminda Dharmaratne,1,329,622 Simon I Hay,1,329 Ali H Mokdad,1,329 David M Pigott,1,329 Robert C Reiner,1,329 Theo Vos1,329

ABSTRACT been identified over the past three decades in population health Background While there is a long history of measuring death and research have included obtaining data in data-sparse,­ burden-­ disability from injuries, modern research methods must account for heavy areas of the world, developing adjustments for ill-defined­ the wide spectrum of disability that can occur in an injury, and must causes of death, separately estimating cause of injury from the provide estimates with sufficient demographic, geographical and bodily harm that results from an injury event and adjusting for temporal detail to be useful for policy makers. The Global Burden of known biases in data, such as underestimation in sexual violence Disease (GBD) 2017 study used methods to provide highly detailed data.3 8 9 Cumulatively, the global injuries research community estimates of global injury burden that meet these criteria. has developed a wide array of methodological innovations and Methods In this study, we report and discuss the methods used in advancements to overcome many of these challenges, although GBD 2017 for injury morbidity and mortality burden estimation. In undoubtedly the science will continue to advance as higher-­ summary, these methods included estimating cause-­specific mortality quality datasets become available, as modelling methods improve for every cause of injury, and then estimating incidence for every cause and as computational processing power becomes more accessible of injury. Non-­fatal disability for each cause is then calculated based to population health research groups around the world. on the probabilities of suffering from different types of bodily injury Many studies have been published based on different releases experienced. of the GBD study, ranging from studies on intentional injuries in Results GBD 2017 produced morbidity and mortality estimates for the eastern Mediterranean to detailed assessments of traumatic 38 causes of injury. Estimates were produced in terms of incidence, brain injury and spinal cord injury disability rates on a global prevalence, years lived with disability, cause-­specific mortality, years scale.10 11 While this array of published GBD injury studies of life lost and disability-­adjusted life-­years for a 28-­year period for 22 demonstrates a broad spectrum of expert knowledge on specific age groups, 195 countries and both sexes. injuries or specific geographies or both, it is also critical to recog- Conclusions GBD 2017 demonstrated a complex and sophisticated nise that population health is a rapidly evolving, collaborative series of analytical steps using the largest known database of morbidity science that has benefited from near-continual­ improvements and mortality data on injuries. GBD 2017 results should be used to even through the current updates being implemented for GBD help inform injury prevention policy making and resource allocation. We 2019. As a result, it should benefit the scientific enterprise to also identify important avenues for improving injury burden estimation focus on publishing the most updated results with perspective on in the future. global, demographic and temporal patterns, and on sharing iter- ative updates on the current state of the science of GBD injuries burden estimation. The goal of this study is to comprehensively review and report methods used for GBD 2017 and associated http://injuryprevention.bmj.com/ INTRODUCTION publications that have gone through extensive collaborator-­ The Global Burden of Disease (GBD) study is a comprehen- review and peer-review­ processes. sive assessment of population health loss. GBD has expanded in scope since its original release in 1994 (GBD 1990) and was most recently updated in autumn 2018 (GBD 2017).1–7 Each METHODS update of the study has provided updated results through the GBD 2017 study most recent year of data availability as well as increasingly GBD is predicated on the principle that every case of death and refined detail in terms of locations, age groups and causes. In disability in the population should be systematically identified addition, GBD incorporates new data as well as updated methods and accounted for in the formulation of global disease and injury for each annual release that represent the expanding complexity burden. On the side of mortality, every death that occurs in the

of the study. Cumulatively, the increasing volume of data and population should have one underlying cause of death which on September 27, 2021 by guest. Protected copyright. increasingly sophisticated estimation methods have necessitated can be assigned to a cause in a mutually exclusive, collectively near-continual­ refinements in terms of data processing, statis- exhaustive hierarchy of diseases and injuries that can cause tical modelling, computational storage and processing as well as death. These data can be used in a method described below to global collaboration with the over 4000 GBD collaborators in calculate cause-specific­ mortality rates and years of life lost. For over 140 countries and territories. morbidity, every non-­fatal case of disease or injury should have Historically, injuries have formed one of the three broad cause an amount of disability assigned for some period of time. These groups in the GBD cause hierarchy alongside the other two main data can be used in a process described below to estimate the groups of health loss (communicable, maternal, neonatal and incidence, prevalence and years lived with disability. Summing nutritional diseases; non-­communicable diseases). Not surpris- morbidity and mortality from some cause form the burden from ingly, there is considerable variation in how morbidity and that cause, expressed as disability-adjusted­ life-­years (DALY). For mortality are estimated across different causes in the GBD hier- causes with known risk factors, some portion of this burden may archy and study design. The methods for estimating morbidity be explained by exposure to that risk factor. Across causes within and mortality from injuries have evolved over time through the some population, it is also a principle of GBD that the sum of most recent release of GBD 2017. Historically, there have been all cause-specific­ deaths should equal all-cause­ mortality in the certain challenges in injuries burden estimation, some of which population, and that rates of incidence, prevalence, remission have been addressed and updated over time, and some of which and cause-specific­ mortality can be reconciled with one another remain as methodological challenges to address as population such that all death and disability in a population is internally health measurement develops more sophisticated modelling consistent across causes and geographies. As examples, the sum strategies. For example, methodological challenges that have of different types of road injury cases must sum up to overall i128 James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research road injuries, and the sum of deaths from different injuries Table 1 Global Burden of Disease cause-­of-­injury hierarchy in a given country must sum up to the estimate of all-­injury deaths. The principle of internal consistency extends to popu- lations used in GBD, where every birth, death and net migra- Forces of nature, tion must be accounted for in the population estimates which Self-­harm and conflict and terrorism Unintentional interpersonal and executions and form the denominators of GBD results. While there is immense Transport injuries injuries violence police conflict complexity in the process summarised above, it is important to Road injuries Falls Self-harm­ Exposure to forces of begin with these core principles which govern the computation nature processes at the heart of GBD burden estimation. A summarised Pedestrian road Drowning Self-­harm by firearm Conflict and terrorism overview of key GBD 2017 methods is also provided in online injuries Cyclist road injuries Fire, heat and hot Self-­harm by other Executions and police supplementary appendix 1. substances specified means conflict Motorcyclist road Poisonings Interpersonal violence GBD study design and hierarchies injuries Motor vehicle road Poisoning by carbon Assault by firearm GBD study design, including cause-specific­ methods, is described injuries monoxide 2–7 in a high level of detail in associated publications. In addi- Other road injuries Poisoning by other Assault by sharp object tion to the injury-­focused methods described in this paper, it is means important to define hierarchies used in the GBD study design. Other transport Exposure to Assault by other means In particular, GBD 2017 was built around a location hierarchy injuries mechanical forces Unintentional firearm where different subnational locations (eg, US states, India states, injuries China provinces) which form a composite of a national location Unintentional (eg, the USA, India, China). National locations are aggregated suffocation to form GBD regions, which are then aggregated to form GBD Other exposure to super regions. These designations affect the modelling structure mechanical forces and utilisation of location random effects, processes which are Adverse effects of medical treatment described in more detail later. The country-­level and regional-­ Animal contact level GBD location hierarchy used in GBD 2017 is provided in Venomous animal online supplementary appendix table 1. In addition to locations, contact GBD processes are conducted to produce estimates for every Non-venomous­ one of 22 age groups, male and female sex and across 28 years animal contact from 1990 to 2017 (inclusive). Age-standardised,­ all-age­ and Foreign body combined sex results are also computed for each GBD result. Pulmonary aspiration and foreign body in Exceptions exist to the rules above, for example, self-harm­ is not airway permitted to occur in the 0–6 days (early neonatal) age group

Foreign body in eyes http://injuryprevention.bmj.com/ in the GBD age hierarchy. There are no sex restrictions placed Foreign body in other on any GBD injury causes, although these restrictions exist for body part other GBD causes, such as cancers like prostate, cervical and Environmental heat uterine being related to one sex. and cold exposure Other unintentional injuries GBD injury classification In the GBD cause hierarchy, injuries are part of the first level of disabling than a mild skin abrasion, it is important for measuring the GBD cause hierarchy, which consists of three broad groups: non-fatal­ burden to consider both the cause and the nature in communicable, maternal, neonatal and nutritional diseases; the formulation of complete injury burden. A full list of nature non-communicable­ diseases and injuries. Additional levels of the of injury is provided in table 3.

GBD cause hierarchy provide additional detail. The hierarchy on September 27, 2021 by guest. Protected copyright. of injuries in GBD is provided in table 1. The organisation of Cause-specific mortality and years of life lost the hierarchy has implications both in terms of how results are As described above, cause-specific­ mortality is measured for every produced and in terms of analytical and processing steps which cause of injury in the GBD cause hierarchy with the exception of are discussed in more detail below. Case definitions including foreign body in the ear and sexual violence, which undergo only International Classification of Diseases (ICD) codes used to non-fatal­ burden estimation (described in more detail below). identify injury deaths and cases are provided in table 2. GBD adheres to five general principles for measuring cause-­ GBD separates the concept of cause of injury from nature of specific mortality, which are described in more detail elsewhere injury. Cause of injury (eg, road injuries, falls, drowning) have but are summarised as follows.12 First, GBD 2017 identifies all historically been used for assigning cause of death as opposed to available data. For injuries, this includes vital registration (VR), the ‘nature’ of injury, which more directly specifies the pathology vital registration samples, verbal autopsy (VA), police records that resulted in death. For example, an individual who falls, and mortuary/hospital data. VR is the preferred data source but fractures his or her hip, undergoes surgery and then develops is not available in every location in the GBD location hierarchy. hospital-­acquired pneumonia and dies while hospitalised would Prior VA research has demonstrated that VA is more accurate still have a fall as the underlying cause of death, regardless of for certain injury causes than it is for certain diseases.13 Police whether sepsis or some other disease process leads to death data undergo additional validity checks to ensure that systematic more proximally in the chain of events. In this individual, the under-­reporting does not occur in comparison to VR data, which ‘nature’ of injury would have been specified as a hip fracture, is described in more detail in a related publication.6 The second since it is the bodily injury that would dictate the disability this general principle relevant to injury mortality estimation is maxi- person experiences. Since it is evident that a hip fracture is more mising comparability and quality of the dataset. For the purposes

James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 i129 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research

Table 2 Case definitions for cause of injury in GBD 2017

Child causes ICD codes Case definition (fatal) Case definition (non-­fatal)

Self-harm­ ICD9: E950-­E959 Deliberate bodily damage inflicted on oneself resulting in Deliberate bodily damage inflicted on oneself with or without ICD10: X60-­X64.9, X66-­X84.9, Y87.0 death intent to kill oneself. Self-harm­ by firearm ICD9: E955-­E955.9 Deliberate bodily damage inflicted by firearm on oneself Deliberate bodily damage inflicted on oneself by firearm with or ICD10: X72-­X74.9 resulting in death without intent to kill oneself. Self-­harm by other specified means ICD9: E950-­E954, E956-­E958.0, E958.2-­E959 Deliberate bodily damage inflicted on oneself resulting in Deliberate bodily damage inflicted on oneself with or without ICD10: X60-­X64.9, X66-X67.9,­ X69-­X71.9, X75- death by means of:* intent to kill oneself by means of:* X75.9, X77-­X84.9, Y87.0 ►► Self-poisoning­ ►► Self-poisoning­ ►► Medication overdose ►► Medication overdose ►► Transport incident ►► Transport incident ►► Falling from height ►► Falling from height ►► Hanging/strangulation ►► Hanging/strangulation *(not exhaustive) *(not exhaustive) Poisoning ICD9: E850.3-­E858.99, E862-­E869.99, E929.2 Death resulting from accidental exposure to a non-infectious­ Unintentional exposure to a non-­infectious substance which ICD10: J70.5, X40-X44.9,­ X47-X49.9,­ Y10-Y14.9,­ substance which contacts the body or enters into the body contacts the body or enters into the body via inhalation, Y16-Y19.9­ via inhalation, ingestion, injection or absorption and causes ingestion, injection or absorption and causes deranged deranged physiological function of body and/or cellular physiological function of body and/or cellular injury/death. injury/death. Poisoning by carbon monoxide (CO) ICD9: E862-­E862.99, E868-­E869.99 Death from exposure to carbon monoxide (CO) as identified Non-­fatal exposure to CO as identified based on ICD10: J70.5, X47-­X47.9 based on carboxyhemoglobin levels (specified based on carboxyhemoglobin levels (specified based on smoking status smoking status and age) or proximity to a confirmed CO and age) or proximity to a confirmed CO poisoning case. poisoning case.

Poisoning by other means ICD9: E850.3-­E858.99, E866-­E866.99 Death resulting from accidental exposure to a non-infectious­ Accidental exposure to a non-­infectious substance (other ICD10: X40-­X44.9, X49-­X49.9, Y10-­Y14.9, substance (other than CO) which contacts the body or than CO) which contacts the body or enters into the body Y16-Y19.9­ enters into the body via inhalation, ingestion, injection or via inhalation, ingestion, injection or absorption and causes absorption and causes deranged physiological function of deranged physiological function of body and/or cellular injury/ body and/or cellular injury/death. death. Animal contact ICD9: E905-­E906.99 Death resulting from unintentionally being attacked, struck, Bodily damage resulting from unintentionally being attacked, ICD10: W52.0-­W62.9, W64-­W64.9, X20-­X29.9 impaled, bitten, stung, crushed, exposed to or stepped on by butted, impaled, bitten, stung, crushed, exposed to or stepped a non-­human animal. on by a non-human­ animal. Venomous animal contact ICD9: E905-­E905.99 Death resulting from unintentionally being bitten by, stung Bodily damage resulting from unintentionally being bitten by, ICD10: W52.3, X20-X29.9­ by, or exposed to a non-human­ venomous animal. stung by or exposed to a non-­human venomous or poisonous animal. Non-venomous­ animal contact ICD9: E905-­E906.99 Death resulting from unintentionally being attacked, struck, Bodily damage resulting from unintentionally being attacked, ICD10: W52.0-­W62.9, W64-­W64.9, X20-­X29.9 impaled, crushed, exposed to or stepped on by a non-­human struck, impaled, crushed, exposed to or stepped on by a non-­ animal. human animal. Falls ICD9: E880-­E886.99, E888-­E888.9, E929.3 A sudden movement downwards due to slipping, tripping or A sudden movement downward due to slipping, tripping or ICD10: W00-­W19.9 other accidental movement which results in a person coming other accidental movement which results in a person coming to rest inadvertently on the ground, floor or other lower to rest inadvertently on the ground, floor or other lower level, level, resulting in death. resulting in tissue damage. http://injuryprevention.bmj.com/ Drowning ICD10: W65-­W70.9, W73-­W74.9 Death that occurs as a result of immersion in water or Non-fatal­ immersion or submersion in water or another fluid, ICD9: E910-­E910.99 another fluid. regardless of whether tissue damage has occurred. The subject can be resuscitated and has not suffered brain death. Fire, heat, and hot substances ICD9: E890-­E899.09, E924-­E924.99, E929.4 Death due to unintentional exposure to substances of high Unintentional exposure to substances of high temperature ICD10: X00-­X06.9, X08-­X19.9 temperature sufficient to cause tissue damage on exposure, sufficient to cause tissue damage on exposure, including bodily including bodily contact with hot liquid, solid or gas such as contact with hot liquid, solid or gas such as cooking stoves, cooking stoves, smoke, steam, drinks, machinery, appliances, smoke, steam, drinks, machinery, appliances, tools, radiators and tools, radiators and objects radiating heat energy. objects radiating heat energy. Road injuries ICD9: E800.3, E801.3, E802.3, E803.3, E804.3, Interaction with an automobile, motorcycle, pedal cycle or Interaction with an automobile, motorcycle, pedal cycle or other E805.3, E806.3, E807.3, E810.0-­E810.6, other vehicles resulting in death. vehicles resulting in bodily damage. E811.0-E811.7,­ E812.0-E812.7,­ E813.0- E813.7, E814.0-­E814.7, E815.0-­E815.7, E816.0-E816.7,­ E817.0-E817.7,­ E818.0- E818.7, E819.0-­E819.7, E820.0-­E820.6, E821.0-E821.6,­ E822.0-E822.7,­ E823.0- E823.7, E824.0-­E824.7, E825.0-­E825.7, E826.0-E826.1,­ E826.3-E826.4,­ E827.0, on September 27, 2021 by guest. Protected copyright. E827.3-­E827.4, E828.0, E828.4, E829.0- E829.4 ICD10: V01-­V04.99, V06-­V80.929, V82-­V82.9, V87.2-V87.3­ Pedestrian road injuries ICD9: E811.7, E812.7, E813.7, E814.7, E815.7, Interaction, as a pedestrian on the road, with an automobile, Interaction, as a pedestrian on the road, with an automobile, E816.7, E817.7, E818.7, E819.7, E822.7, motorcycle, pedal cycle or other vehicles resulting in death. motorcycle, pedal cycle or other vehicles resulting in bodily E823.7, E824.7, E825.7, E826.0, E827.0, damage. E828.0, E829.0 ICD10: V01-­V04.99, V06-­V09.9 Cyclist road injuries ICD9: E800.3, E801.3, E802.3, E803.3, E804.3, Accident, as a cyclist or passenger on a pedal cycle, resulting Accident, as a cyclist or passenger on a pedal cycle, resulting in E805.3, E806.3, E807.3, E810.6, E811.6, in death. bodily damage. E812.6, E813.6, E814.6, E815.6, E816.6, E817.6, E818.6, E819.6, E820.6, E821.6, E822.6, E823.6, E824.6, E825.6, E826.1 ICD10: V10-­V19.9

Motorcyclist road injuries ICD9: E810.2-E810.3,­ E811.2-E811.3,­ E812.2- Accident, as a rider on a motorcycle, resulting in death. Accident, as a rider on a motorcycle, resulting in bodily damage. E812.3, E813.2-­E813.3, E814.2-­E814.3, E815.2-E815.3,­ E816.2-E816.3,­ E817.2- E817.3, E818.2-­E818.3, E819.2-­E819.3, E820.2-E820.3,­ E821.2-E821.3,­ E822.2- E822.3, E823.2-­E823.3, E824.2-­E824.3, E825.2-E825.3­ ICD10: V20-­V29.9

Continued i130 James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research

Table 2 Continued

Child causes ICD codes Case definition (fatal) Case definition (non-­fatal)

Motor vehicle road injuries ICD9: E810.0-­E810.1, E811.0-­E811.1, E812.0- Accident, as a driver or passenger in a motor vehicle, Accident, as a driver or passenger in a motor vehicle, resulting E812.1, E813.0-­E813.1, E814.0-­E814.1, resulting in death. in bodily damage. E815.0-E815.1,­ E816.0-E816.1,­ E817.0- E817.1, E818.0-­E818.1, E819.0-­E819.1, E820.0-E820.1,­ E821.0-E821.1,­ E822.0- E822.1, E823.0-­E823.1, E824.0-­E824.1, E825.0-E825.1­ ICD10: V30-­V79.9, V87.2-­V87.3 Other road injuries ICD9: E810.4-­E810.5, E811.4-­E811.5, E812.4- Death resulting from being a driver or passenger of a Bodily damage resulting from being a driver or passenger of E812.5, E813.4-­E813.5, E814.4-­E814.5, vehicle not including automobiles, motorcycles, bicycles (ie, a vehicle not including automobiles, motorcycles, bicycles (ie, E815.4-E815.5,­ E816.4-E816.5,­ E817.4- streetcar). streetcar). E817.5, E818.4-­E818.5, E819.4-­E819.5, E820.4-E820.5,­ E821.4-E821.5,­ E822.4- E822.5, E823.4-­E823.5, E824.4-­E824.5, E825.4-E825.5,­ E826.3-E826.4,­ E827.3- E827.4, E828.4, E829.4 ICD10: V80-­V80.929, V82-­V82.9 Other transport injuries ICD9: E800-­E800.2, E801-­E801.2, E802-­E802.2, Interaction with a means of transport other than automobile, Interaction with a means of transport other than automobile, E803-E803.2,­ E804-E804.2,­ E805-E805.2,­ motorcycle, pedal cycle or other road vehicles resulting in motorcycle, pedal cycle or other road vehicles resulting in bodily E806-E806.2,­ E807-E807.2,­ E810.7, death. damage. E820.7, E821.7, E826.2, E827.2, E828.2, E830-E838.9,­ E840-E849.9,­ E929.1 ICD10: V00-­V00.898, V05-­V05.99, V81-­V81.9, V83-V86.99,­ V88.2-V88.3,­ V90-V98.8­ Interpersonal violence ICD9: E960-­E969 Death from intentional use of physical force or power, Sustaining bodily harm in terms of tissue damage from ICD10: X85-­Y08.9, Y87.1-­Y87.2 threatened or actual, from another person or group not intentional use of physical force or power, threatened or actual, including military or police forces. from another person or group not including military or police forces. Physical violence by firearm ICD9: E965-­E965.4 Death from intentional use of physical force or power by Sustaining bodily harm in terms of tissue damage from ICD10: X93-­X95.9 a firearm from another person or group or community not intentional use of physical force or power by a firearm from including military or police forces. another person or group not including military or police forces. Physical violence by sharp object ICD9: E966 Death from intentional use of physical force or power by a Sustaining bodily harm in terms of tissue damage from ICD10: X99-­X99.9 sharp object from another person or group or community not intentional use of physical force or power by a sharp object from including military or police forces. another person or group not including military or police forces. Sexual violence ICD9: E960-­E960.1 NA Experiencing at least one event of sexual violence in the last ICD10: Y05-­Y05.9 year, where sexual violence is defined as any sexual assault, including both penetrative sexual violence (rape) and non-­ penetrative sexual violence (other forms of unwanted sexual touching). Physical violence by other means ICD9: E961-­E964, E965.5-­E965.9, E967-­E969 Death from intentional use of physical force or power by Sustaining bodily harm in terms of tissue damage from ICD10: X85-­X92.9, X96-­X98.9, Y00-­Y04.9, Y06- an object other than a firearm or sharp object from another intentional use of physical force or power by an object other Y08.9, Y87.1-­Y87.2 person or group or community not including military or than a firearm or sharp object from another person or group not police forces. including military or police forces. http://injuryprevention.bmj.com/ Conflict and terrorism ICD9: E979-­E979.9, E990-­E999.1 Death resulting from the instrumental use of violence by Bodily harm resulting from the instrumental use of violence ICD10: U00-­U03, Y36-­Y38.9, Y89.1 people who identify themselves as members of a group— by people who identify themselves as members of a group— whether this group is transitory or has a more permanent whether this group is transitory or has a more permanent identity—against another group or set of individuals, in identity—against another group or set of individuals, in order to order to achieve political, economic or social objectives. achieve political, economic or social objectives. Executions and police conflict ICD9: E970-­E978 State-­sanctioned executions or police-­related altercations State-­sanctioned executions or police-related­ altercations ICD10: Y35-Y35.93,­ Y89.0 leading to death. leading to bodily damage. Exposure to forces of nature ICD9: E907-­E909.9 Death resulting from an unforeseen and often sudden Bodily damage resulting from an unforeseen and often sudden ICD10: X33-­X38.9 natural event such as a hurricane, earthquake, tsunami or natural event such as a hurricane, earthquake, tsunami or tornado. tornado. Exposure to mechanical forces ICD9: E913-­E913.19, E916-­E922.99, E928.1- Unintentional death resulting from contact with or threat of Unintentional bodily damage resulting from contact with or E928.7 an (in)animate object, human or plant. threat of an (in)animate object, human or plant. ICD10: W20-­W38.9, W40-­W43.9, W45.0-­W45.2, W46-W46.2,­ W49-W52,­ W75-W76.9­ Unintentional firearm injuries ICD9: E922-­E922.99, E928.7 Unintentional death resulting from contact with a firearm. Unintentional bodily damage resulting from contact with a ICD10: W32-­W34.9 firearm. on September 27, 2021 by guest. Protected copyright. Other exposure to mechanical forces ICD9: E916-­E921.99, E928.1-­E928.6 Unintentional death resulting from contact with or threat Unintentional bodily damage resulting from contact with or ICD10: W20-­W31.9, W35-­W38.9, W40-­W43.9, of an (in)animate object (not including a firearm), human threat of an (in)animate object (not including a firearm), human W45.0-W45.2,­ W46-W46.2,­ W49-W52­ or plant. or plant. Pulmonary aspiration and foreign body ICD9: 770.1–770.18, E911-­E912.09, E913.8- Unintentional death from inhaling, swallowing or aspirating Unintentional bodily damage from inhaling, swallowing or in airway E913.99 extraneous materials or substance that enters the airway aspirating extraneous materials or substance that enters the ICD10: W78-­W80.9, W83-­W84.9 or lungs. airway or lungs. Foreign body in eyes ICD9: 360.5–360.69, 374.86, 376.6, E914- NA Unintentional damage from extraneous materials or substance E914 09 in the orbital structure or eye. ICD10: H02.81-­H02.819, H44.6-­H44.799 Foreign body in other body part ICD9: 709.4, E915-­E915.09 Unintentional death from an extraneous material or Unintentional bodily damage from an extraneous material or ICD10: M60.2-­M60.28, W44-­W45, W45.3-­W45.9 substance being within the body, not including the airway, substance being within the body, not including the airway, lungs or eyes. lungs or eyes.

Injuries definition: damage, defined by cellular death, tissue disruption, loss of homeostasis, pain limiting activities of daily living or short-term­ psychological harm (for cases of sexual violence), inflicted on the body as the direct or indirect result of a physical force, immersion or exposure, which may include interpersonal or self-inflicted­ forces. GBD, Global Burden of Disease; ICD, International Classification of Diseases.

of injury mortality estimation, this process is largely focused detail elsewhere) and (3) processing VA studies into usable data on (1) ensuring appropriate accounting for different ICD code that map to the GBD cause hierarchy.8 9 12 The third general prin- versions used for cause of death data classification over time, (2) ciple for injury cause of death models in GBD 2017 is to develop redistribution of ill-defined­ causes of death (described in more a diverse set of plausible models. This process is conducted via

James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 i131 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research

Table 3 GBD nature of injury Nature of injury Amputation of lower limbs, bilateral Fracture of sternum and/or fracture of one or more ribs Crush injury Amputation of upper limbs, bilateral Fracture of vertebral column Nerve injury Amputation of fingers (excluding thumb) Fracture of femur, other than femoral neck Injury to eyes Amputation of lower limb, unilateral Minor TBI Poisoning requiring urgent care Amputation of upper limb, unilateral Moderate/severe TBI Severe chest injury Amputation of thumb Spinal cord lesion at neck level Internal haemorrhage in abdomen and pelvis Amputation of toe/toes Spinal cord lesion below neck level Effect of different environmental factors Lower airway burns Muscle and tendon injuries, including sprains and strains Complications following therapeutic procedures lesser dislocations Burns, <20% total burned surface area without lower airway burns Foreign body in ear Multiple fractures, dislocations, crashes, wounds, pains and strains Burns, ≥20% total burned surface area or ≥10% burned surface area if head/neck Open wound(s) or hands/wrist involved without lower airway burns Fracture of clavicle, scapula or humerus Contusion in any part of the body Fracture of face bones Superficial injury of any part of the body Fracture of foot bones except ankle Dislocation of hip Fracture of hand (wrist and other distal part of hand) Dislocation of knee Fracture of hip Dislocation of shoulder Fracture of patella, tibia or fibula or ankle Foreign body in respiratory system Fracture of pelvis Foreign body in GI and urogenital system Fracture of radius and/or ulna Drowning and non-­fatal submersion Fracture of skull Asphyxiation GBD, Global Burden of Disease; GI, gastrointestinal; TBI, traumatic brain injury. the Cause of Death Ensemble model (CODEm) framework, vital registration data (Colombia, females) is shown in figure 1. which is the standard, peer-reviewed­ cause of death estimation A similar model with relatively less data is shown in figure 2 process used extensively in the GBD study. CODEm generates (Honduras, females). While data are absent in more recent years a large set of possible models based on covariates suggested by in Honduras, the model is still able to follow temporal trends, age the modeller based on expert input and literature review (eg, patterns and broader geographical patterns by harnessing signals alcohol for road injuries) and then runs every plausible model, from covariate-based­ fixed effects (eg, alcohol consumption per which can range into the thousands per cause. These models can capita) and location-based­ random effects (eg, the regional trends http://injuryprevention.bmj.com/ be conducted in both rate space and cause fraction space and use in Central Latin America and patterns in neighbouring coun- an assortment of combinations among the user-­selected covari- tries). All cause of death models from GBD 2017 are publicly ates (table 4). Fourth, the predictive validity of each one of these available for review (https://​vizhub.​healthdata.​org/​cod/). Cause-­ submodels is tested using test-­train holdouts, whereby a specific specific deaths are converted to cause-specific­ mortality rates model is trained on a portion of data and tested on a separate (CSMRs) using GBD populations. Once CSMRs are established, portion to determine out-­of-­sample predictive validity. Once the years of life lost (YLLs) are computed as the product of CSMRs submodels are conducted and predictive validity is measured, and residual life expectancy at the age of death. The residual life then an ensemble model is developed out of the submodels. The expectancy is based on the lowest observed mortality rate for submodels and the ensemble model are then subject to the fifth each age across all populations over 5 million. For example, if principle, which is to choose the best-performing­ models based a death from road injuries occurs at age 25 and the residual life

on out-­of-­sample predictive validity. The chosen models may be expectancy is 60 years, then there are 60 YLLs attributed to that on September 27, 2021 by guest. Protected copyright. a single cause model or an ensemble of models. Beyond these death. If the death had occurred at age 50 with a residual life processes, which have become automated with expert review in expectancy of 38 years, then 38 YLLs would be attributed. Life the GBD processing architecture, there is also considerable time tables used for GBD 2017 are provided in related publications. 7 required by the analysts, modellers, collaborators and principal investigators who are involved in the GBD study. Such processes also come under expert scrutiny via the GBD Scientific Council Injury incidence, prevalence and years lived with disability and the peer-review­ process in the annual GBD capstone After cause-specific­ models for each cause of injury in the GBD publications.2–7 cause hierarchy are conducted, the non-­fatal estimation process Once submodels and ensemble models have been conducted is conducted. An overview of this process is depicted in figure 3. for each cause in the GBD cause hierarchy, a process to correct In the first stage, we estimate the incidence of injuries warranting for cause of death rates to ensure internal consistency is medical care using DisMod-MR­ 2.1 (abbreviated DisMod). conducted. Specifically, each subcause within some overall cause DisMod is a meta-regression­ tool for epidemiological estimation is rescaled such that, for example, every subtype of road inju- that uses a compartmental model structure whereby a healthy ries sums to road injuries deaths overall, and then road injuries population may become diseased or injured, at which point the and other transport injuries sum to equal the overall transport individual either remains a prevalent case, goes into remission or injuries cause. As this cascades to the overall cause hierarchy dies. DisMod essentially fits differential equations to reconcile and the overall all-­cause mortality rates, cause-specific­ mortality the transitions between these different compartments, so that across all causes ultimately equals the overall mortality in the the final posterior estimate for each epidemiological param- population. An example of an injuries cause of death model with eter can be explained in the context of the other parameters. i132 James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research

Table 4 Covariates used in GBD cause of death models Cause Global or data-­rich model Sex Number of Covariates used covariates used Transport injuries Global/Data rich Male 10 Alcohol (litres per capita), Education (years per capita), Lag distributed income per capita (I$), Population Density (300–500 ppl/sqkm, proportion), Population Density (500–1000 ppl/sqkm, proportion), Rainfall Quintile 5 (proportion), Vehicles–two+four wheels (per capita), Vehicles–two wheels fraction (proportion), Sociodemographic Index, Healthcare Access and Quality Index Transport injuries Global/Data rich Female 10 Alcohol (litres per capita), Education (years per capita), Lag distributed income per capita (I$), Population Density (300–500 ppl/sqkm, proportion), Population Density (500–1000 ppl/sqkm, proportion), Rainfall Quintile 5 (proportion), Vehicles–two+four wheels (per capita), Vehicles–two wheels fraction (proportion), Sociodemographic Index, Healthcare Access and Quality Index Road injuries Global/Data rich Male 13 Alcohol (liters per capita), Education (years per capita), Lag distributed income per capita (I$), Population 15 to 30 (proportion), Population Density (300–500 ppl/sqkm, proportion), Population Density (500–1000 ppl/sqkm, proportion), Vehicles–two+four wheels (per capita), Vehicles–two wheels (per capita), Vehicles - 4 wheels (per capita), Vehicles–two wheels fraction (proportion), Log-­transformed summary exposure value (SEV) scalar: Road Inj, Sociodemographic Index, Healthcare Access and Quality Index Road injuries Global/Data rich Female 13 Alcohol (liters per capita), Education (years per capita), Lag distributed income per capita (I$), Population 15 to 30 (proportion), Population Density (300–500 ppl/sqkm, proportion), Population Density (500–1000 ppl/sqkm, proportion), Vehicles–two+four wheels (per capita), Vehicles–two wheels (per capita), Vehicles - 4 wheels (per capita), Vehicles–two wheels fraction (proportion), Log-­transformed SEV scalar: Road Inj, Sociodemographic Index, Healthcare access and quality index Pedestrian road Global/Data rich Male 11 Alcohol (liters per capita), Education (years per capita), Population Density (300–500 ppl/sqkm, proportion), injuries Population Density (500–1000 ppl/sqkm, proportion), Rainfall Quintile 5 (proportion), Vehicles–two+four wheels (per capita), Vehicles–two wheels fraction (proportion), Log-­transformed SEV scalar: Pedest, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Pedestrian road Global/Data rich Female 11 Alcohol (liters per capita), Education (years per capita), Population Density (300–500 ppl/sqkm, proportion), injuries Population Density (500–1000 ppl/sqkm, proportion), Rainfall Quintile 5 (proportion), Vehicles–two+four wheels (per capita), Vehicles–two wheels fraction (proportion), Log-­transformed SEV scalar: Pedest, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Cyclist road injuries Global/Data rich Male 10 Alcohol (liters per capita), Education (years per capita), Population Density (300–500 ppl/sqkm, proportion), Population Density (500–1000 ppl/sqkm, proportion), Vehicles–two+four wheels (per capita), Vehicles - two wheels fraction (proportion), Log-­transformed SEV scalar: Cyclist, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Cyclist road injuries Global/Data rich Female 10 Alcohol (liters per capita), Education (years per capita), Population Density (300–500 ppl/sqkm, proportion), Population Density (500–1000 ppl/sqkm, proportion), Vehicles - two+four wheels (per capita), Vehicles–two wheels fraction (proportion), Log-­transformed SEV scalar: Cyclist, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Motorcyclist road Global/Data rich Male 10 Alcohol (liters per capita), Education (years per capita), Population Density (300–500 ppl/sqkm, proportion), injuries Population Density (500–1000 ppl/sqkm, proportion), Rainfall Quintile 5 (proportion), Vehicles–two wheels (per capita), Log-­transformed SEV scalar: Mot Cyc, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$)

Motorcyclist road Global/Data rich Female 10 Alcohol (liters per capita), Education (years per capita), Population Density (300–500 ppl/sqkm, proportion), http://injuryprevention.bmj.com/ injuries Population Density (500–1000 ppl/sqkm, proportion), Rainfall Quintile 5 (proportion), Vehicles–two wheels (per capita), Log-­transformed SEV scalar: Mot Cyc, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Motor vehicle road Global/Data rich Male 10 Alcohol (liters per capita), Education (years per capita), Population Density (300–500 ppl/sqkm, proportion), injuries Population Density (500–1000 ppl/sqkm, proportion), Rainfall Quintile 5 (proportion), Vehicles–four wheels (per capita), Log-­transformed SEV scalar: Mot Veh, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Motor vehicle road Global/Data rich Female 10 Alcohol (liters per capita), Education (years per capita), Population Density (300–500 ppl/sqkm, proportion), injuries Population Density (500–1000 ppl/sqkm, proportion), Rainfall Quintile 5 (proportion), Vehicles–four wheels (per capita), Log-­transformed SEV scalar: Mot Veh, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Other road injuries Global/Data rich Male 8 Alcohol (liters per capita), Rainfall Quintile 5 (proportion), Vehicles–two+four wheels (per capita), Vehicles–two wheels fraction (proportion), Log-­transformed SEV scalar: Oth Road, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) on September 27, 2021 by guest. Protected copyright. Other road injuries Global/Data rich Female 8 Alcohol (liters per capita), Rainfall Quintile 5 (proportion), Vehicles–two+four wheels (per capita), Vehicles–two wheels fraction (proportion), Log-­transformed SEV scalar: Oth Road, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Other transport Global/Data rich Male 11 Alcohol (liters per capita), Education (years per capita), Population Density (300–500 ppl/sqkm, proportion), injuries Population Density (500–1000 ppl/sqkm, proportion), Rainfall Quintile 5 (proportion), Vehicles–two+four wheels (per capita), Vehicles–two wheels fraction (proportion), Log-­transformed SEV scalar: Oth Trans, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Other transport Global/Data rich Female 11 Alcohol (liters per capita), Education (years per capita), Population Density (300–500 ppl/sqkm, proportion), injuries Population Density (500–1000 ppl/sqkm, proportion), Rainfall Quintile 5 (proportion), Vehicles–two+four wheels (per capita), Vehicles–two wheels fraction (proportion), Log-­transformed SEV scalar: Oth Trans, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Falls Global/Data rich Male 7 Alcohol (liters per capita), Elevation Over 1500 m (proportion), Log-transformed­ SEV scalar: Falls, Sociodemographic Index, milk adjusted(g), Healthcare Access and Quality Index, Lag distributed income per capita (I$) Falls Global/Data rich Female 7 Alcohol (liters per capita), Elevation Over 1500 m (proportion), Log-transformed­ SEV scalar: Falls, Sociodemographic Index, milk adjusted(g), Healthcare Access and Quality Index, Lag distributed income per capita (I$) Drowning Global/Data rich Male 10 Alcohol (liters per capita), Coastal Population within 10 km (proportion), Education (years per capita), Landlocked Nation (binary), Elevation Under 100 m (proportion), Rainfall Quintile 1 (proportion), Rainfall Quintile 5 (proportion), Log-­transformed SEV scalar: Drown, Sociodemographic Index, Lag distributed income per capita (I$) Drowning Global/Data rich Female 10 Alcohol (liters per capita), Coastal Population within 10 km (proportion), Education (years per capita), Landlocked Nation (binary), Elevation Under 100 m (proportion), Rainfall Quintile 1 (proportion), Rainfall Quintile 5 (proportion), Log-­transformed SEV scalar: Drown, Sociodemographic Index, Lag distributed income per capita (I$)

Continued James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 i133 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research

Table 4 Continued

Fire, heat and hot Global/Data rich Male 9 Alcohol (liters per capita), Tobacco (cigarettes per capita), Education (years per capita), Indoor Air Pollution substances (All Cooking Fuels), Population Density (over 1000 ppl/sqkm, proportion), Log-transformed­ SEV scalar: Fire, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Fire, heat and hot Global/Data rich Female 9 Alcohol (liters per capita), Tobacco (cigarettes per capita), Education (years per capita), Indoor Air Pollution substances (All Cooking Fuels), Population Density (over 1000 ppl/sqkm, proportion), Log-transformed­ SEV scalar: Fire, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Poisonings Global/Data rich Male 8 Education (years per capita), Opium Cultivation (binary), Population Density (over 1000 ppl/sqkm, proportion), Population Density (under 150 ppl/sqkm, proportion), Log-­transformed SEV scalar: Poison, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Poisonings Global/Data rich Female 8 Education (years per capita), Opium Cultivation (binary), Population Density (over 1000 ppl/sqkm, proportion), Population Density (under 150 ppl/sqkm, proportion), Log-­transformed SEV scalar: Poison, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Poisoning by carbon Global/Data rich Male 4 Education (years per capita), Lag distributed income per capita (I$), Sociodemographic Index, Healthcare Access monoxide and Quality Index Poisoning by carbon Global/Data rich Female 4 Education (years per capita), Lag distributed income per capita (I$), Sociodemographic Index, Healthcare access monoxide and quality index Poisoning by other Global/Data rich Male 4 Education (years per capita), Lag distributed income per capita (I$), Sociodemographic Index, Healthcare access means and quality index Poisoning by other Global/Data rich Female 4 Education (years per capita), Lag distributed income per capita (I$), Sociodemographic Index, Healthcare access means and quality index Exposure to Global/Data rich Male 7 Alcohol (liters per capita), Education (years per capita), Population Density (over 1000 ppl/sqkm, proportion), mechanical forces Population Density (under 150 ppl/sqkm, proportion), Sociodemographic Index, Healthcare access and quality index, Lag distributed income per capita (I$) Exposure to Global/Data rich Female 7 Alcohol (liters per capita), Education (years per capita), Population Density (over 1000 ppl/sqkm, proportion), mechanical forces Population Density (under 150 ppl/sqkm, proportion), Sociodemographic Index, Healthcare access and quality index, Lag distributed income per capita (I$) Unintentional firearm Global/Data rich Male 9 Alcohol (liters per capita), Education (years per capita), Health System Access (unitless), Population Density (over injuries 1000 ppl/sqkm, proportion), Population Density (under 150 ppl/sqkm, proportion), Log-transformed­ SEV scalar: Mech Gun, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Unintentional firearm Global/Data rich Female 9 Alcohol (liters per capita), Education (years per capita), Health System Access (unitless), Population Density (over injuries 1000 ppl/sqkm, proportion), Population Density (under 150 ppl/sqkm, proportion), Log-transformed­ SEV scalar: Mech Gun, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Other exposure to Global/Data rich Male 9 Alcohol (liters per capita), Education (years per capita), Health System Access (unitless), Population Density (over mechanical forces 1000 ppl/sqkm, proportion), Population Density (under 150 ppl/sqkm, proportion), Log-transformed­ SEV scalar: Oth Mech, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Other exposure to Global/Data rich Female 9 Alcohol (liters per capita), Education (years per capita), Health System Access (unitless), Population Density (over mechanical forces 1000 ppl/sqkm, proportion), Population Density (under 150 ppl/sqkm, proportion), Log-transformed­ SEV scalar: Oth Mech, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$)

Adverse effects of Global/Data rich Male 3 Lag distributed income per capita (I$), Sociodemographic Index, Healthcare Access and Quality Index http://injuryprevention.bmj.com/ medical treatment Adverse effects of Global/Data rich Female 3 Lag distributed income per capita (I$), Sociodemographic Index, Healthcare Access and Quality Index medical treatment Animal contact Global/Data rich Male 11 Alcohol (liters per capita), Education (years per capita), Elevation Over 1500 m (proportion), Population 15 to 30 (proportion), Population Density (over 1000 ppl/sqkm, proportion), Population Density (under 150 ppl/sqkm, proportion), Elevation Under 100 m (proportion), Log-­transformed SEV scalar: Animal, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Animal contact Global/Data rich Female 11 Alcohol (liters per capita), Education (years per capita), Elevation Over 1500 m (proportion), Population 15 to 30 (proportion), Population Density (over 1000 ppl/sqkm, proportion), Population Density (under 150 ppl/sqkm, proportion), Elevation Under 100 m (proportion), Log-­transformed SEV scalar: Animal, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Venomous animal Global/Data rich Male 10 Alcohol (liters per capita), Education (years per capita), Elevation Over 1500 m (proportion), Population Density contact (over 1000 ppl/sqkm, proportion), Population Density (under 150 ppl/sqkm, proportion), Elevation Under 100 m (proportion), Log-­transformed SEV scalar: Venom, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) on September 27, 2021 by guest. Protected copyright. Venomous animal Global/Data rich Female 10 Alcohol (liters per capita), Education (years per capita), Elevation Over 1500 m (proportion), Population Density contact (over 1000 ppl/sqkm, proportion), Population Density (under 150 ppl/sqkm, proportion), Elevation Under 100 m (proportion), Log-­transformed SEV scalar: Venom, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Non-venomous­ animal Global Male 6 Alcohol (liters per capita), Education (years per capita), Lag distributed income per capita (I$), Log-transformed­ contact SEV scalar: Non Ven, Sociodemographic Index, Healthcare Access and Quality Index Non-venomous­ animal Data rich Male 10 Alcohol (liters per capita), Education (years per capita), Elevation Over 1500 m (proportion), Population Density contact (over 1000 ppl/sqkm, proportion), Population Density (under 150 ppl/sqkm, proportion), Elevation Under 100 m (proportion), Log-­transformed SEV scalar: Non Ven, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Non-venomous­ animal Global/Data rich Female 10 Alcohol (liters per capita), Education (years per capita), Elevation Over 1500 m (proportion), Population Density contact (over 1000 ppl/sqkm, proportion), Population Density (under 150 ppl/sqkm, proportion), Elevation Under 100 m (proportion), Log-­transformed SEV scalar: Non Ven, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Foreign body Global Male 10 Education (years per capita), Indoor Air Pollution (All Cooking Fuels), Population Density (over 1000 ppl/sqkm, proportion), Population Over 65 (proportion), Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Foreign body Global Female 10 Education (years per capita), Indoor Air Pollution (All Cooking Fuels), Population Density (over 1000 ppl/sqkm, proportion), Population Over 65 (proportion), Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$)

Continued i134 James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research

Table 4 Continued

Pulmonary aspiration Global/Data rich Male 6 Alcohol (liters per capita), Lag distributed income per capita (I$), Mean BMI, Log-transformed­ SEV scalar: F Body and foreign body in Aspn, Sociodemographic Index, Access and Quality Index airway Pulmonary aspiration Global Female 8 Alcohol (liters per capita), Education (years per capita), Mean BMI, Alcohol binge drinker proportion, age-­ and foreign body in standardised, Log-­transformed SEV scalar: F Body Aspn, Sociodemographic Index, Healthcare access and quality airway index, Lag distributed income per capita (I$) Pulmonary aspiration Data rich Female 6 Alcohol (liters per capita), Lag distributed income per capita (I$), Mean BMI, Log-transformed­ SEV scalar: F Body and foreign body in Aspn, Sociodemographic Index, Healthcare Access and Quality Index airway Foreign body in other Global/Data rich Male 10 Alcohol (liters per capita), Education (years per capita), Elevation Over 1500 m (proportion), Population Density body part (over 1000 ppl/sqkm, proportion), Population Density (under 150 ppl/sqkm, proportion), Elevation Under 100 m (proportion), Log-­transformed SEV scalar: Oth F Body, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Foreign body in other Global/Data rich Female 10 Alcohol (liters per capita), Education (years per capita), Elevation Over 1500 m (proportion), Population Density body part (over 1000 ppl/sqkm, proportion), Population Density (under 150 ppl/sqkm, proportion), Elevation Under 100 m (proportion), Log-­transformed SEV scalar: Oth F Body, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Environmental heat Global/Data rich Male 11 Education (years per capita), Lag distributed income per capita (I$), Population-weighted­ mean temperature, and cold exposure Elevation Over 1500 m (proportion), Elevation 500 to 1500 m (proportion), Population Density (150–300 ppl/sqkm, proportion), Rainfall (Quintiles 4–5), Sanitation (proportion with access), 90th percentile climatic temperature in the given country-­year, Sociodemographic Index, Healthcare Access and Quality Index Environmental heat Global/Data rich Female 11 Education (years per capita), Lag distributed income per capita (I$), Population-weighted­ mean temperature, and cold exposure Elevation Over 1500 m (proportion), Elevation 500 to 1500 m (proportion), Population Density (150–300 ppl/sqkm, proportion), Rainfall (Quintiles 4–5), Sanitation (proportion with access), 90th percentile climatic temperature in the given country-­year, Sociodemographic Index, Healthcare Access and Quality Index Other unintentional Global/Data rich Male 12 Alcohol (liters per capita), Education (years per capita), Elevation Over 1500 m (proportion), Population Density injuries (over 1000 ppl/sqkm, proportion), Population Density (under 150 ppl/sqkm, proportion), Elevation Under 100 m (proportion), Vehicles–two wheels (per capita), Vehicles–four wheels (per capita), Log-transformed­ SEV scalar: Oth Unint, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Other unintentional Global/Data rich Female 12 Alcohol (liters per capita), Education (years per capita), Elevation Over 1500 m (proportion), Population Density injuries (over 1000 ppl/sqkm, proportion), Population Density (under 150 ppl/sqkm, proportion), Elevation Under 100 m (proportion), Vehicles–two wheels (per capita), Vehicles–four wheels (per capita), Log-transformed­ SEV scalar: Oth Unint, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Self-harm­ Global Male 11 Alcohol (liters per capita), Education (years per capita), Population Density (150–300 ppl/sqkm, proportion), Population Density (300–500 ppl/sqkm, proportion), Population Density (500–1000 ppl/sqkm, proportion), Population Density (over 1000 ppl/sqkm, proportion), Population Density (under 150 ppl/sqkm, proportion), Sociodemographic Index, Healthcare Access and Quality Index, Muslim Religion (proportion of population), Lag distributed income per capita (I$) Self-harm­ Global Female 15 Alcohol (liters per capita), Education (years per capita), Population Density (150–300 ppl/sqkm, proportion), Population Density (300–500 ppl/sqkm, proportion), Population Density (500–1000 ppl/sqkm, proportion),

Population Density (over 1000 ppl/sqkm, proportion), Population Density (under 150 ppl/sqkm, proportion), http://injuryprevention.bmj.com/ Religion (binary,>50% Muslim), Log-transformed­ SEV scalar: Self Harm, Sociodemographic Index, Major depressive disorder, Risk of selfharm due to major depressive disorder, Healthcare Access and Quality Index, Non-­ partner lifetime prevalence of sexual violence (female-only),­ Lag distributed income per capita (I$) Self-harm­ Data rich Male 11 Alcohol (liters per capita), Education (years per capita), Population Density (150–300 ppl/sqkm, proportion), Population Density (300–500 ppl/sqkm, proportion), Population Density (500–1000 ppl/sqkm, proportion), Population Density (over 1000 ppl/sqkm, proportion), Population Density (under 150 ppl/sqkm, proportion), Religion (binary,>50% Muslim), Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Self-harm­ Data rich Female 13 Alcohol (liters per capita), Education (years per capita), Population Density (150–300 ppl/sqkm, proportion), Population Density (300–500 ppl/sqkm, proportion), Population Density (500–1000 ppl/sqkm, proportion), Population Density (over 1000 ppl/sqkm, proportion), Population Density (under 150 ppl/sqkm, proportion), Religion (binary,>50% Muslim), Log-transformed­ SEV scalar: Self Harm, Sociodemographic Index, Major depressive disorder, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Self-­harm by firearm Global/Data rich Male 13 Alcohol (liters per capita), Education (years per capita), Population Density (150–300 ppl/sqkm, proportion), Population Density (300–500 ppl/sqkm, proportion), Population Density (500–1000 ppl/sqkm, proportion), on September 27, 2021 by guest. Protected copyright. Population Density (over 1000 ppl/sqkm, proportion), Population Density (under 150 ppl/sqkm, proportion), Religion (binary,>50% Muslim), Log-transformed­ SEV scalar: Self Harm, Sociodemographic Index, Major depressive disorder, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Self-­harm by firearm Global/Data rich Female 13 Alcohol (liters per capita), Education (years per capita), Population Density (150–300 ppl/sqkm, proportion), Population Density (300–500 ppl/sqkm, proportion), Population Density (500–1000 ppl/sqkm, proportion), Population Density (over 1000 ppl/sqkm, proportion), Population Density (under 150 ppl/sqkm, proportion), Religion (binary,>50% Muslim), Log-transformed­ SEV scalar: Self Harm, Sociodemographic Index, Major depressive disorder, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Self-­harm by other Global/Data rich Male 13 Alcohol (liters per capita), Education (years per capita), Population Density (150–300 ppl/sqkm, proportion), specified means Population Density (300–500 ppl/sqkm, proportion), Population Density (500–1000 ppl/sqkm, proportion), Population Density (over 1000 ppl/sqkm, proportion), Population Density (under 150 ppl/sqkm, proportion), Religion (binary,>50% Muslim), Log-transformed­ SEV scalar: Self Harm, Sociodemographic Index, Major depressive disorder, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Self-­harm by other Global/Data rich Female 13 Alcohol (liters per capita), Education (years per capita), Population Density (150–300 ppl/sqkm, proportion), specified means Population Density (300–500 ppl/sqkm, proportion), Population Density (500–1000 ppl/sqkm, proportion), Population Density (over 1000 ppl/sqkm, proportion), Population Density (under 150 ppl/sqkm, proportion), Religion (binary,>50% Muslim), Log-transformed­ SEV scalar: Self Harm, Sociodemographic Index, Major depressive disorder, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Interpersonal violence Global/Data rich Male 8 Alcohol (liters per capita), Education (years per capita), Opium Cultivation (binary), Population Density (over 1000 ppl/sqkm, proportion), Log-­transformed SEV scalar: Violence, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$)

Continued

James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 i135 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research

Table 4 Continued

Interpersonal violence Global/Data rich Female 8 Alcohol (liters per capita), Education (years per capita), Opium Cultivation (binary), Population Density (over 1000 ppl/sqkm, proportion), Log-­transformed SEV scalar: Violence, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Physical violence by Global/Data rich Male 8 Alcohol (liters per capita), Education (years per capita), Opium Cultivation (binary), Population Density (over firearm 1000 ppl/sqkm, proportion), Log-­transformed SEV scalar: Viol Gun, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Physical violence by Global/Data rich Female 8 Alcohol (liters per capita), Education (years per capita), Opium Cultivation (binary), Population Density (over firearm 1000 ppl/sqkm, proportion), Log-­transformed SEV scalar: Viol Gun, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Physical violence by Global/Data rich Male 8 Alcohol (liters per capita), Education (years per capita), Opium Cultivation (binary), Population Density (over sharp object 1000 ppl/sqkm, proportion), Log-­transformed SEV scalar: Viol Knife, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Physical violence by Global/Data rich Female 8 Alcohol (liters per capita), Education (years per capita), Opium Cultivation (binary), Population Density (over sharp object 1000 ppl/sqkm, proportion), Log-­transformed SEV scalar: Viol Knife, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Physical violence by Global/Data rich Male 8 Alcohol (liters per capita), Education (years per capita), Opium Cultivation (binary), Population Density (over other means 1000 ppl/sqkm, proportion), Log-­transformed SEV scalar: Oth Viol, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Physical violence by Global/Data rich Female 8 Alcohol (liters per capita), Education (years per capita), Opium Cultivation (binary), Population Density (over other means 1000 ppl/sqkm, proportion), Log-­transformed SEV scalar: Oth Viol, Sociodemographic Index, Healthcare Access and Quality Index, Lag distributed income per capita (I$) Executions and police Global/Data rich Male 6 Alcohol (liters per capita), Education (years per capita), Lag distributed income per capita (I$), Population Density conflict (over 1000 ppl/sqkm, proportion), Sociodemographic Index, Healthcare Access and Quality Index Executions and police Global/Data rich Female 6 Alcohol (liters per capita), Education (years per capita), Lag distributed income per capita (I$), Population Density conflict (over 1000 ppl/sqkm, proportion), Sociodemographic Index, Healthcare Access and Quality Index

BMI, body mass index.

Similar to the principles described in CODEm, DisMod uses definition as we do not want to consider minor stumbles and all available data, ranging from incidence data to cause-specific­ falls, for example, that led to no actual bodily harm as injuries for mortality rates from the corrected CODEm results, to produce GBD, since they would not have any associated disability. These estimates for every age, sex, year and location. For the purposes models are conducted only for injury causes as opposed to the of injuries, we established our case definition for non-fatal­ inju- nature of injuries references above. Each data input is designated ries as injuries that require medical care. This is a necessary case based on type of data—specifically, inpatient data, outpatient http://injuryprevention.bmj.com/ on September 27, 2021 by guest. Protected copyright.

Figure 1 Cause of Death Ensemble model with data points for road injuries in Colombia for females.

i136 James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research

Figure 2 Cause of Death Ensemble model with data points for road injuries in Honduras for females

data, surveillance data, survey data and literature studies that are used for adjustment are provided in table 5. Input data for injury http://injuryprevention.bmj.com/ population-­representative. We model incidence rates for hospital cause incidence models included sources identified as part of admissions for injuries, so the non-­inpatient data sources get systematic reviews conducted in past GBD cycles, new sources adjusted according to their classification so that the model inputs identified by the GBD collaborator network and new sources of are consistent as injuries that warranted or received inpatient clinical data and other injuries data obtained by the core injuries medical care. The coefficients measured by DisMod that were burden estimation team at the Institute for Health Metrics and on September 27, 2021 by guest. Protected copyright.

Figure 3 Injuries non-­fatal estimation flow chart.

James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 i137 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research

Table 5 Covariates and coefficients used in Global Burden of Disease incidence cause models Cause Outpatient coefficient Injury receiving formal care, inpatient Injury warranting medical care and outpatient coefficient coefficient Animal contact 7.04 (7.03–7.04) 7.56 (6.91–8.31) Non-­venomous animal contact 2.91 (2.91–2.92) 11.21 (10.1–12.38) Venomous animal contact 3.14 (3.01–3.34) 4.09 (3.69–4.5) Drowning 0.88 (0.87–0.89) 1.01 (1.0–1.05) 30.42 (15.33–51.11) Falls 6.91 (6.89–6.94) 5.94 (5.5–6.46) 9.73 (9.28–10.22) Fire, heat and hot substances 3.53 (3.53–3.56) 7.82 (7.24–8.51) Pulmonary aspiration and foreign body in airway 3.37 (3.35–3.43) 15.36 (13.93–16.86) Foreign body in eyes 931.4 (923.34–934.49) 302.06 (251.14–365.04) Foreign body in other body part 1.97 (1.95–2.01) 20.97 (15.55–26.26) Interpersonal violence 6.57 (6.56–6.61) 21.43 (13.6–32.79) 46.97 (39.57–53.62) Assault by firearm 1.36 (1.29–1.44) 1.27 (1.05–1.6) 53.58 (50.65–54.54) Assault by sharp object 3.18 (2.92–3.5) 2.38 (1.86–3.22) 37.91 (28.3–50.05) Assault by other means 5.65 (5.44–5.89) 2.44 (2.02–3.2) Exposure to mechanical forces 12.4 (12.0–12.82) 33.3 (30.51–36.23) Unintentional firearm injuries 2.71 (2.53–2.9) 4.6 (3.49–6.36) Other exposure to mechanical forces 12.62 (12.55–12.85) 30.77 (25.74–36.09) Adverse effects of medical treatment 1.06 (1.06–1.06) 19.81 (17.29–26.1) Environmental heat and cold exposure 3.91 (3.9–3.94) 17.54 (3.91–49.6) Other unintentional injuries 13.53 (13.46–13.78) 14.95 (9.62–24.12) Poisonings 3.96 (3.73–4.19) 3.78 (3.4–4.21) 8.47 (4.41–16.64) Poisoning by carbon monoxide 5.86 (5.68–5.92) Poisoning by other means 4.18 (3.9–4.5) Self-harm­ 2.75 (2.75–2.78) 2.5 (2.2–2.83) Self-­harm by firearm 2.77 (2.42–3.07) 16.94 (2.81–51.06) Self-­harm by other specified means 1.5 (1.47–1.51) 6.73 (2.78–19.14) Other transport injuries 1.65 (1.6–1.77) 1.01 (1.0–1.03) Road injuries 3.77 (3.75–3.78) 6.16 (5.65–6.68) 15.44 (13.25–18.1) Motorcyclist road injuries 1.94 (1.92–1.99) Motor vehicle road injuries 4.48 (4.46–4.48) http://injuryprevention.bmj.com/ Other road injuries 6.9 (6.89–6.96) Cyclist road injuries 4.54 (4.33–4.89) Pedestrian road injuries 1.94 (1.94–1.96) 15.78 (7.63–36.6)

Evaluation at the University of Washington. In addition, CSMRs cause. The proportion of new cases of injury that would have from the corrected CODEm models described above are used some nature of injury as the most disabling outcome is deter- in this stage of DisMod modelling. The list of non-­fatal injury mined based on dual-coded­ clinical data sources where both sources used in GBD 2017 is provided in online supplementary the cause and nature of injury were included as ICD codes.10

appendix table 2. The completed DisMod models for inpatient Of note, one limitation of this process is that due to computa- on September 27, 2021 by guest. Protected copyright. incidence for each cause of injury are publicly available at https://​ tional demands, it is currently only possible to apportion the vizhub.​healthdata.​org/​epi/. most disabling nature of injury for each new case of injury. Once an incidence cause model is constructed for each cause of As such, the probability that each nature of injury is the most injury, an extensive analytical ‘pipeline’ follows which converts disabling nature of injury for some cause of injury is modelled injury cause incidence into years lived with disability. First, in a Dirichlet regression such that the probabilities sum to 1. In inpatient incidence is split into inpatient and outpatient inci- other words, each nature of injury has some probability of being dence using coefficients empirically measured by DisMod. The the most disabling injury suffered by the victim of some cause of outpatient coefficients for each injury cause are also included injury, but if multiple natures of injury occurred, then the less in table 5. Separate pipelines are then conducted for inpatient disabling injuries are not captured as part of that injury cause’s and outpatient injury incidence—each step below can be consid- disability. This limitation has been recognised as a limitation of ered to have been run for both streams of data, for each cause GBD injury burden estimation in various peer-­reviewed articles of injury. After the coefficient is applied, incidence is adjusted and will likely be addressed in future GBD updates as computa- by the excess mortality rate measured by DisMod to essentially tional efficiency improves.3 10 remove injury cases that died after the injury occurred. Once The probability distributions of each cause-nature­ are these deaths are removed from the incidence pool, the resulting computed separately for each age, sex, year and location. At this steps are applied to these surviving cases of injury. First, each point, the analytical stage has the age-­specific, sex-specific,­ year-­ new case of injury is considered to have 47 possible ‘natures’ specific, location-specific­ incidence of a cause-nature­ combina- of injury that can result. These are the types of bodily injury tion, for example, the incidence of road injuries that led to a that are considered to be possible outcomes from a given injury cervical-level­ spinal cord injury in males aged 20–24 years in i138 James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research

2017 in Stockholm, Sweden. The next step converts these inci- Table 6 Short-­term disability weights for each nature of injury dence estimates into short-­term and long-­term injury incidence estimates, where long-term­ disability is defined as having a lower Short-­term disability functional status 1 year postinjury than at the time of injury. Nature of injury weight These probabilities were measured using long-term­ follow-up­ Spinal cord lesion at neck level 0.7319 studies.14–20 For some natures of injury, such as lower extremity Spinal cord lesion below neck level 0.6235 amputation, the probability of being a long-term­ injury is 1. Foreign body in respiratory system 0.4079 The probabilities of short-­term versus long-­term injury for Lower airway burns 0.3764 each cause-­nature combination are used to split the incidence Severe chest Injury 0.3685 values into short-­term and long-­term pipelines. The long-­term Internal haemorrhage in abdomen and pelvis 0.3242 incidence is then converted to prevalence using the ordinary Burns, ≥20% total burned surface area or ≥10% burned 0.3145 differential equation solver used in DisMod, which also uses as surface area if head/neck or hands/wrist involved without an input excess mortality estimated for certain natures of injury lower airway burns such as traumatic brain injury and spinal cord injury conducted Fracture of pelvis 0.2788 in a previous systematic review and meta-­analysis. The short-­ Fracture of hip 0.2575 term incidence is converted to prevalence by multiplying inci- Multiple fractures, dislocations, crashes, wounds, sprains 0.2575 dence and duration of injury, where duration of injury was either and strains computed directly from follow-­up studies or, in the case of Drowning and non-­fatal submersion 0.2471 unavailable data, estimated by an expert clinical panel involved Asphyxiation 0.2471 in previous iterations of the GBD study. Since access to medical Moderate TBI 0.2137 treatment is assumed to affect duration of injury and disability, Poisoning requiring urgent care 0.1628 the GBD Healthcare Access and Quality Index is used to estimate Burns, <20% total burned surface area without lower 0.1408 the proportion with and without access to medical treatment on airway burns 21 a location-specific­ basis. The average duration for short-term­ Effect of different environmental factors 0.1334 injury is therefore calculated as the percentage treated multiplied Complications following therapeutic procedures 0.1334 by treated duration added to the percentage untreated multiplied Crush injury 0.1325 by the untreated duration. The output from this step is the short-­ Foreign body in GI and urogenital system 0.1143 term prevalence of each cause-nature­ combination. Short-term­ Dislocation of knee 0.1134 prevalence is subtracted from long-term­ prevalence at this stage Fracture of femur, other than femoral neck 0.1114 to avoid double counting the same case of injury. Once short-­ Fracture of vertebral column 0.1106 term and long-­term prevalence estimates for each cause-­nature Minor TBI 0.11 are computed, then disability weights as derived by the Salomon Fracture of sternum and/or fracture of one or more ribs 0.1027 et al process are assigned to each injury nature.22 Short-­term disability weights by injury nature are shown in table 6, which Nerve injury 0.0997 http://injuryprevention.bmj.com/ does not include amputations since we assume they cause only Fracture of skull 0.0714 long-term­ disability. The full list of long-­term disability weights Fracture of face bones 0.0669 by injury nature, location and year are provided in online Dislocation of shoulder 0.062 supplementary appendix table 3, which does not include foreign Injury to eyes 0.0543 body in respiratory system, foreign body in gastrointestinal and Fracture of patella, tibia or fibula or ankle 0.0501 urogenital system, foreign body in ear and superficial injury of Fracture of clavicle, scapula or humerus 0.0349 any part of body, since we assume these natures of injury do not Fracture of radius and/or ulna 0.0281 cause long-term­ disability. After disability weights are assigned Fracture of foot bones except ankle 0.026 to each injury case, years lived with disability for each cause of Dislocation of hip 0.0159 injury are calculated as the prevalence of each health state multi- Foreign body in ear 0.0133

plied by the corresponding disability weight and then summed Fracture of hand (wrist and other distal part of hand) 0.0099 on September 27, 2021 by guest. Protected copyright. across natures of injury for each cause to compute years lived Muscle and tendon injuries, including sprains and strains 0.0075 with disability (YLDs) for each age, sex, year and location for lesser dislocations that injury cause. YLDs then undergo comorbidity adjustment Contusion in any part of the body 0.0075 used across the GBD study whereby comorbid cases of disease Superficial injury of any part of the body 0.0075 and injury in the population are simulated and adjusted disability Open wound(s) 0.0058 weights are computed. These processes are described in more GI, gastrointestinal; TBI, traumatic brain injury. detail in GBD literature.3 GBD 2017 provided an important methodological update whereby nature of injury results, regard- less of cause of injury, could be reviewed in the results from this rates in the GBD 2017 study to estimate the yearly propor- process; this has enabled more advanced GBD research such as tion of the population that experienced at least one episode measuring the burden of traumatic brain injury and spinal cord of sexual violence in the past year, using a case definition of injury, measuring the burden of facial fractures and measuring any sexual assault including penetrative sexual violence (rape) the burden of hand and finger fractures.10 and non-­penetrative sexual violence (other forms of unwanted sexual touching). To inform the sexual violence estimates, we identified data in 93 countries that met the case definition Sexual violence above. This resulted in 263 site-­years of data, which mainly Sexual violence follows a different analytical pathway than the were derived from surveys such as Demographic and Health other causes of injury. This process is shown in figure 4. We used Surveys and Reproductive Health Surveys. Similar to our other the same study framework as was developed for other injury injury models, we used DisMod 2.1 to model prevalence. The

James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 i139 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research

Figure 4 Sexual violence estimation flow chart. HAQI, Healthcare Access and Quality Index. sexual violence prevalence model used study-­level covariates be related to more dynamic temporal trends in injury mortality for each type of survey question, for example, we used a study-­ patterns over time. In general, most data points exist within level covariate to identify surveys that identify penetrative the 95% UI of the model fit (mean: 98.5% in-sample, 98.0% sexual violence only to account for how the overall incidence of out-­of-­sample). sexual violence is greater than this value. This model also used a covariate on alcohol use in litres per capita for each location to Incidence models help fit the model in data-­sparse locations. Once yearly preva- Model performance metrics for each injury cause model in GBD lence was measured, sexual violence cases undergo a process by 2017 are provided in table 8. These model performance metrics which short-­term disability from the physical and psychological include in-sample­ coverage and RMSE of estimated results for harm of sexual violence cases is assigned to each prevalent case; cause-specific­ mortality, excess mortality and incidence. There however, long-­term sequelae of sexual violence are currently not are no performance metrics for CSMR or excess mortality for captured in this process, which has been a known limitation of foreign body in eyes since we do not estimate mortality from this sexual violence estimation in the GBD framework. cause of injury. For incidence, the in-­sample coverage average was 55.3% across cause-of­ -­injury models and ranged from a Disability-adjusted life-years low of 26% in falls to a high of 88% in poisoning by carbon After estimation of cause-­specific mortality and YLLs as well as monoxide. Incidence RMSE ranged from a low of 1.04 in pedes- non-­fatal health outcomes estimation including YLDs, DALYs are trian road injuries to a high of 4.86 in foreign body in eye. calculated as the sum of YLLs and YLDs for each cause of injury. http://injuryprevention.bmj.com/ YLDs are also calculated for each nature of injury category. DISCUSSION GATHER statement Many considerable advancements have been made in the GBD 2017 adheres to the Guidelines for Accurate and Trans- measurement of global injury burden since early versions of the parent Health Estimates Reporting (GATHER). GATHER is GBD Study. Novel datasets, sophisticated statistical modelling described in more detail in online supplementary appendix 2. and global collaboration have all facilitated the advancement of injury burden measurement science. Many more advancements RESULTS in future updates should be possible as larger datasets become available and as computational power allows for more detailed Results for all GBD 2017 injury estimates are available in associ- measurement processes. Continued global collaboration will be ated publications as well as online. Specifically, results by age, sex,

an integral component. Suggested priority items for the advance- on September 27, 2021 by guest. Protected copyright. year, subnational location and nature of injury can be viewed and ment of injury burden estimation are as follows: downloaded online via the GBD Results Tool (http://​ghdx.​health- First, while much of the global injury burden occurs in low-­ data.​org/​gbd-​results-​tool) and GBD Compare (https://​vizhub.​ income and middle-income­ countries, these countries are healthdata.org/​ ​gbd-​compare/). These results are available in terms frequently the most data-­sparse. GBD has rigorously attempted of incidence, prevalence, YLDs, cause-specific­ mortality, YLLs and to collect all available data, including police records and verbal DALYs, expressed in counts, rates, and percentages. Analytical autopsy studies and inpatient and outpatient records; however, code and input datasets are available at http://ghdx.​ ​healthdata.org.​ it is likely that additional data sources in data-sparse­ countries CODEm models exist. Parties who are aware of additional data sources that Model performance metrics for each injury cause model in GBD could be used in the GBD estimation framework should consider 2017 are provided in table 7. Model performance metrics for joining the GBD collaborator network to contribute new sources CODEm models include root mean square error (RMSE) for of data to be used in future estimation updates. in-sample­ tests and out-­of-­sample tests, percentage of data points Second, computational and data limitations make it difficult that correctly predict the trend in-sample­ and out-of­ -­sample to account for the full disability that might be experienced in the and percentage of data points that are present within the 95% setting of multiple injuries. For example, if an individual sustains a uncertainty intervals (UIs) of the model fit. RMSE in-­sample is below-neck­ spinal injury and an upper extremity amputation, the generally better than RMSE out-­of-­sample, which is an expected amputation is not directly accounted for in the prevalence or YLD result that also demonstrates the importance of performing estimate of the injury cause to which this disability is attributed. out-of­ -­sample predictive validity tests. While the correct trend This problem quickly grows in complexity, as one can imagine an is predicted in approximately one in five models, this may also event like a road injury leading to multiple contusions and abrasions, i140 James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research

Table 7 Performance metrics for each cause-­of-­injury CODEm model Per cent coverage Per cent coverage Cause Type Sex RMSE in-­sample RMSE out-­of-­sample in-­sample out-­of-­sample Transport injuries Data rich Female 0.153062 0.211028 0.999851 0.999395 Transport injuries Data rich Male 0.144423 0.202366 0.99978 0.998995 Transport injuries Global Female 0.216405 0.338398 0.99951 0.992996 Transport injuries Global Male 0.209561 0.327954 0.999347 0.99108 Road injuries Data rich Female 0.154916 0.22011 0.999945 0.999642 Road injuries Data rich Male 0.147432 0.208989 0.99987 0.999452 Road injuries Global Female 0.198002 0.338885 0.999736 0.993674 Road injuries Global Male 0.193896 0.321219 0.999332 0.990834 Pedestrian road injuries Data rich Female 0.183693 0.327964 0.999776 0.998965 Pedestrian road injuries Data rich Male 0.177994 0.323544 0.999688 0.998913 Pedestrian road injuries Global Female 0.240151 0.430127 0.999174 0.992328 Pedestrian road injuries Global Male 0.247329 0.409191 0.998229 0.990017 Cyclist road injuries Data rich Female 0.219965 0.435983 0.999892 0.999106 Cyclist road injuries Data rich Male 0.206919 0.500591 0.999876 0.999158 Cyclist road injuries Global Female 0.296895 0.528063 0.998384 0.990875 Cyclist road injuries Global Male 0.294776 0.527441 0.998702 0.988234 Motorcyclist road Data rich Female 0.268406 0.653692 0.999776 0.998805 injuries Motorcyclist road Data rich Male 0.195368 0.444714 0.999793 0.998395 injuries Motorcyclist road Global Female 0.362655 0.692762 0.998726 0.99082 injuries Motorcyclist road Global Male 0.283024 0.502588 0.998804 0.987794 injuries Motor vehicle road Data rich Female 0.167766 0.33083 0.99993 0.999335 injuries Motor vehicle road Data rich Male 0.160584 0.309726 0.999919 0.999377 injuries Motor vehicle road Global Female 0.230946 0.38664 0.99957 0.995355 injuries http://injuryprevention.bmj.com/ Motor vehicle road Global Male 0.232898 0.378096 0.999353 0.992869 injuries Other road injuries Data rich Female 0.408852 1.04171 0.997205 0.970506 Other road injuries Data rich Male 0.467256 1.21047 0.994429 0.9463 Other road injuries Global Female 0.558784 0.899497 0.994899 0.96375 Other road injuries Global Male 0.654189 1.0708 0.984753 0.931697 Other transport injuries Data rich Female 0.255843 0.406371 0.999581 0.998655 Other transport injuries Data rich Male 0.195575 0.404214 0.999666 0.99863 Other transport injuries Global Female 0.31846 0.546918 0.998599 0.991384 Other transport injuries Global Male 0.267514 0.49731 0.998444 0.989304 Falls Data rich Female 0.162773 0.237492 0.999873 0.999522 on September 27, 2021 by guest. Protected copyright. Falls Data rich Male 0.157114 0.220452 0.999847 0.999492 Falls Global Female 0.246877 0.428822 0.99923 0.988577 Falls Global Male 0.246101 0.369118 0.999571 0.989585 Drowning Data rich Female 0.177905 0.258172 0.999932 0.999782 Drowning Data rich Male 0.164617 0.226899 0.999868 0.999373 Drowning Global Female 0.238598 0.428467 0.999657 0.992777 Drowning Global Male 0.224438 0.361879 0.99961 0.989534 Fire, heat and hot Data rich Female 0.175426 0.245 0.999962 0.999793 substances Fire, heat and hot Data rich Male 0.17054 0.227618 0.999944 0.999737 substances Fire, heat and hot Global Female 0.281428 0.401798 0.999483 0.994548 substances Fire, heat and hot Global Male 0.289708 0.40982 0.999518 0.99422 substances Poisonings Data rich Female 0.190498 0.283924 0.999901 0.999732 Poisonings Data rich Male 0.189747 0.283639 0.999888 0.999668

Continued James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 i141 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research

Table 7 Continued Per cent coverage Per cent coverage Cause Type Sex RMSE in-­sample RMSE out-­of-­sample in-­sample out-­of-­sample Poisonings Global Female 0.311328 0.515718 0.99918 0.993385 Poisonings Global Male 0.323815 0.529806 0.999166 0.992089 Poisoning by carbon Data rich Female 0.255034 0.352342 0.999119 0.998139 monoxide Poisoning by carbon Data rich Male 0.234913 0.328692 0.999486 0.998765 monoxide Poisoning by carbon Global Female 0.353393 0.688269 0.998372 0.982832 monoxide Poisoning by carbon Global Male 0.305615 0.621778 0.999006 0.983458 monoxide Poisoning by other Data rich Female 0.208468 0.470199 0.999861 0.998144 means Poisoning by other Data rich Male 0.231395 0.543185 0.999871 0.998948 means Poisoning by other Global Female 0.284383 0.555132 0.999746 0.989287 means Poisoning by other Global Male 0.288098 0.590913 0.999759 0.990146 means Exposure to mechanical Data rich Female 0.171902 0.29354 0.999636 0.99932 forces Exposure to mechanical Data rich Male 0.162641 0.259268 0.999605 0.998955 forces Exposure to mechanical Global Female 0.398855 0.54379 0.995672 0.987855 forces Exposure to mechanical Global Male 0.325975 0.454021 0.995758 0.985214 forces Unintentional firearm Data rich Female 0.207177 0.502831 0.999619 0.999488 injuries Unintentional firearm Data rich Male 0.221533 0.49235 0.999306 0.998449 injuries

Unintentional firearm Global Female 0.354152 0.591674 0.998979 0.991558 http://injuryprevention.bmj.com/ injuries Unintentional firearm Global Male 0.355798 0.64953 0.996524 0.980841 injuries Other exposure to Data rich Female 0.20287 0.436518 0.999912 0.999795 mechanical forces Other exposure to Data rich Male 0.170292 0.318704 0.999896 0.999761 mechanical forces Other exposure to Global Female 0.406425 0.538089 0.995379 0.98994 mechanical forces Other exposure to Global Male 0.361646 0.472713 0.995528 0.988955 mechanical forces Adverse effects of Data rich Female 0.186809 0.305147 0.999832 0.999511 on September 27, 2021 by guest. Protected copyright. medical treatment Adverse effects of Data rich Male 0.217278 0.342415 0.999833 0.999577 medical treatment Adverse effects of Global Female 0.280204 0.430453 0.999698 0.993818 medical treatment Adverse effects of Global Male 0.277028 0.431272 0.999573 0.992957 medical treatment Animal contact Data rich Female 0.277226 0.439671 0.999355 0.998642 Animal contact Data rich Male 0.231627 0.414921 0.999863 0.999528 Animal contact Global Female 0.401714 0.691306 0.998669 0.987713 Animal contact Global Male 0.316647 0.623446 0.9991 0.99176 Venomous animal Data rich Female 0.417726 0.745234 0.960501 0.956152 contact Venomous animal Data rich Male 0.401006 0.761481 0.977149 0.97478 contact Venomous animal Global Female 0.634642 0.915323 0.965066 0.949503 contact

Continued i142 James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research

Table 7 Continued Per cent coverage Per cent coverage Cause Type Sex RMSE in-­sample RMSE out-­of-­sample in-­sample out-­of-­sample Venomous animal Global Male 0.449848 0.839185 0.97819 0.96024 contact Non-venomous­ animal Data rich Female 0.304776 0.593881 0.994547 0.991865 contact Non-venomous­ animal Data rich Male 0.304223 0.529077 0.998929 0.998113 contact Non-venomous­ animal Global Female 0.421204 0.680417 0.995082 0.9848 contact Non-venomous­ animal Global Male 0.471148 0.740524 0.998707 0.990622 contact Foreign body Data rich Female 0.170699 0.275966 0.999937 0.999705 Foreign body Data rich Male 0.166161 0.263143 0.999798 0.999305 Foreign body Global Female 0.216832 0.401408 0.999535 0.992467 Foreign body Global Male 0.227414 0.381598 0.999262 0.989838 Pulmonary aspiration Data rich Female 0.174424 0.374749 0.999979 0.999572 and foreign body in airway Pulmonary aspiration Data rich Male 0.178947 0.34741 0.999928 0.999294 and foreign body in airway Pulmonary aspiration Global Female 0.267697 0.416038 0.999413 0.993624 and foreign body in airway Pulmonary aspiration Global Male 0.286472 0.422915 0.998089 0.990215 and foreign body in airway Foreign body in other Data rich Female 0.31229 0.664465 0.99005 0.987846 body part Foreign body in other Data rich Male 0.291172 0.629172 0.993547 0.991666 body part

Foreign body in other Global Female 0.462299 0.749894 0.98392 0.971743 http://injuryprevention.bmj.com/ body part Foreign body in other Global Male 0.478614 0.759133 0.984301 0.971436 body part Other unintentional Data rich Female 0.266367 0.450437 0.999612 0.999067 injuries Other unintentional Data rich Male 0.228051 0.387409 0.999597 0.998959 injuries Other unintentional Global Female 0.354782 0.671813 0.997343 0.984969 injuries Other unintentional Global Male 0.301256 0.54085 0.997963 0.985982 injuries

Self-harm­ Data rich Female 0.157456 0.236415 0.999699 0.999206 on September 27, 2021 by guest. Protected copyright. Self-harm­ Data rich Male 0.150967 0.223371 0.999688 0.999011 Self-harm­ Global Female 0.219988 0.370761 0.998551 0.986222 Self-harm­ Global Male 0.203341 0.347213 0.999389 0.979274 Self-­harm by firearm Data rich Female 0.215778 0.439608 0.992476 0.992525 Self-­harm by firearm Data rich Male 0.19323 0.402898 0.998082 0.997457 Self-­harm by firearm Global Female 0.311061 0.642889 0.987894 0.971118 Self-harm­ by firearm Global Male 0.316945 0.590367 0.992646 0.977377 Self-harm­ by other Data rich Female 0.162023 0.345661 0.999855 0.998854 specified means Self-­harm by other Data rich Male 0.235129 0.322581 0.999898 0.999453 specified means Self-harm­ by other Global Female 0.191636 0.38357 0.999636 0.98601 specified means Self-­harm by other Global Male 0.192311 0.348953 0.999813 0.986603 specified means Interpersonal violence Data rich Female 0.224081 0.294307 0.99863 0.996721 Interpersonal violence Data rich Male 0.220852 0.298197 0.998132 0.995665

Continued

James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 i143 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research

Table 7 Continued Per cent coverage Per cent coverage Cause Type Sex RMSE in-­sample RMSE out-­of-­sample in-­sample out-­of-­sample Interpersonal violence Global Female 0.306086 0.450697 0.998456 0.989396 Interpersonal violence Global Male 0.307439 0.479452 0.997588 0.981596 Physical violence by Data rich Female 0.253283 0.414003 0.998598 0.997318 firearm Physical violence by Data rich Male 0.277353 0.501753 0.997843 0.996142 firearm Physical violence by Global Female 0.44617 0.621002 0.993619 0.98712 firearm Physical violence by Global Male 0.41286 0.679294 0.995867 0.981991 firearm Physical violence by Data rich Female 0.222036 0.393235 0.999815 0.999003 sharp object Physical violence by Data rich Male 0.235542 0.463121 0.999796 0.998721 sharp object Physical violence by Global Female 0.276474 0.499795 0.999526 0.993622 sharp object Physical violence by Global Male 0.332336 0.595217 0.999354 0.990212 sharp object Physical violence by Data rich Female 0.204351 0.336239 0.999954 0.999532 other means Physical violence by Data rich Male 0.202192 0.394188 0.999868 0.999051 other means Physical violence by Global Female 0.270287 0.410186 0.999719 0.995718 other means Physical violence by Global Male 0.285589 0.45387 0.999612 0.992595 other means Environmental heat and Data rich Female 0.234754 0.399463 0.999403 0.999073 cold exposure Environmental heat and Data rich Male 0.201821 0.309939 0.999658 0.999207 cold exposure

Environmental heat and Global Female 0.3511 0.639869 0.998595 0.989061 http://injuryprevention.bmj.com/ cold exposure Environmental heat and Global Male 0.33441 0.528137 0.999336 0.993068 cold exposure Executions and police Data rich Female 0.852242 1.4431 0.49803 0.533053 conflict Executions and police Data rich Male 0.970597 1.55607 0.629313 0.628953 conflict Executions and police Global Female 1.2422 1.86518 0.541687 0.549016 conflict Executions and police Global Male 1.04755 1.95756 0.671496 0.659889 conflict on September 27, 2021 by guest. Protected copyright. CODEm, Cause of Death Ensemble model.

several fractures in different anatomical sites, a mild traumatic brain the methods and data required for this update would be complex, injury and a spinal cord injury. There are over 3.6 million permu- they would represent a large increase in the available data that could tations of injury if one considers only 10 possible natures of injury, be used for GBD injuries estimation. making it difficult to quantitatively measure these relationships by Fourth, measuring the total burden of sexual violence has proven cause of injury and by age, sex, year and location. Future research to to be a challenging area of estimation in the GBD framework. As address this limitation may focus on simulation studies that model noted in the ‘Methods’ section of this paper, one known limitation the probability of different comorbid injury combinations to better is how long-term­ sequelae and conditions may not be adequately inform disability weight applications. accounted for in sexual violence burden estimation. In order to Third, more data could be used for nature of injury measure- attribute burden from major depressive disorder, anxiety disorders, ment. Traumatic brain injury and spinal cord injury registries, for self-harm­ and substance use disorders, measuring the relative risk example, are not currently directly compatible with the GBD injury of developing these conditions for victims of sexual violence would estimation framework yet provide rich epidemiological informa- allow for population attributable fractions to be calculated and tion. Future updates to GBD should focus more attention on incor- DALYs from these conditions to be attributed to sexual violence. porating data that measure burden of nature of injury in terms of While the premise of this methodological update is relatively simple, incidence, prevalence or excess mortality. Incorporating these types currently there are relatively few studies to inform these relative of data would require a method to be developed such that estimates risks, and conducting and adding such studies in the future would were internally consistent across cause-nature­ distributions. While be recognised as a major achievement in GBD research as it would i144 James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research

Table 8 Performance metrics for each cause-­of-­injury DisMod model Cause-­specific mortality rate: Cause-­specific mortality Excess mortality rate: Excess mortality rate: Incidence hazard: in-­ Incidence hazard: Cause in-­sample coverage rate: in-­sample RMSE in-­sample coverage in-­sample RMSE sample coverage in-­sample RMSE Animal contact 0.95 0.96 0.69 1.14 0.40 1.64 Non-venomous­ animal contact 0.97 0.98 0.74 1.20 0.53 1.40 Venomous animal contact 0.97 1.13 0.74 1.17 0.48 1.31 Drowning 0.91 0.82 0.84 1.40 0.73 1.61 Falls 0.93 0.66 0.71 1.13 0.26 1.77 Fire, heat and hot substances 0.95 0.59 0.67 0.97 0.50 1.16 Pulmonary aspiration and foreign 0.92 0.93 0.78 1.29 0.65 1.56 body in airway Foreign body in eyes 0.83 4.86 Foreign body in other body part 0.96 1.40 0.74 1.31 0.57 1.39 Interpersonal violence 0.89 0.81 0.64 1.11 0.31 1.77 Assault by firearm 0.93 1.96 0.74 1.07 0.69 1.25 Assault by sharp object 0.92 1.50 0.78 1.05 0.57 1.17 Assault by other means 0.90 0.91 0.75 1.10 0.48 1.33 Exposure to mechanical forces 0.92 0.81 0.61 1.23 0.38 2.01 Unintentional firearm injuries 0.95 1.51 0.75 1.13 0.70 1.17 Other exposure to mechanical 0.93 0.84 0.66 1.22 0.41 1.94 forces Adverse effects of medical 0.92 0.71 0.71 1.48 0.37 1.41 treatment Environmental heat and cold 0.94 1.21 0.73 1.54 0.56 1.52 exposure Other unintentional injuries 0.89 1.31 0.51 1.35 0.50 1.67 Poisonings 0.95 0.90 0.76 1.75 0.58 1.90 Poisoning by carbon monoxide 0.95 0.94 0.81 1.11 0.88 1.17 Poisoning by other means 0.95 0.92 0.79 1.89 0.67 2.04 Self-harm­ 0.98 0.27 0.76 1.02 0.47 1.32 Self-harm­ by firearm 1.00 1.28 0.89 1.31 0.86 1.35 Self-harm­ by other specified 0.98 0.26 0.83 0.96 0.60 1.06 means

Other transport injuries 0.96 0.99 0.73 1.43 0.63 1.32 http://injuryprevention.bmj.com/ Road injuries 0.91 0.47 0.63 1.10 0.27 1.43 Motorcyclist road injuries 0.96 1.07 0.70 1.13 0.54 1.18 Motor vehicle road injuries 0.94 0.55 0.59 1.12 0.48 1.21 Other road injuries 0.99 1.45 0.78 1.16 0.74 1.19 Cyclist road injuries 0.99 1.13 0.73 1.10 0.59 1.09 Pedestrian road injuries 0.92 0.72 0.62 1.02 0.48 1.04 RMSE, root mean square error .

allow for more accurate estimation of lifetime disability caused by For example, a collaborative effort between researchers in sexual violence. This effort would moreover represent an important Vietnam and the Institute for Health Metrics on Evaluation on

contribution to research surrounding the Sustainable Development developing a study on Vietnam injury burden following GBD on September 27, 2021 by guest. Protected copyright. Goals related to sexual violence and women’s rights. 23 24 2017 led to identifying the use of the Vietnam National Injury Fifth, non-­fatal injuries from conflict and natural disaster Survey, which was then added for estimation in GBD 2019. are challenging to estimate because of data sparsity in areas Increasing data collection standardisation efforts should be that are afflicted by these events. Fatalities are estimated after emphasised as a priority in all countries, particularly countries such events, but there is still considerable injury burden among where data coverage on injuries is sparse. Ongoing dialogue the population that survives. Since data collection systems and via scientific publications and international conferences hospitals may also be destroyed in these events, it becomes should also continue to serve as a forum to discuss data and difficult to collect adequate non-fatal­ injury data. Global methodological updates that can continue to refine the science collaboration should also focus on identifying sources of data of injuries estimation in GBD. on non-­fatal and fatal injury cases in conflict and natural disaster events. It will be important to monitor the effects of implementing CONCLUSION these priorities as injury measurement science continues to Measuring injuries burden in GBD is a complex scientific evolve. Global collaborations including the GBD enterprise endeavour that leverages large amounts of data, a complex should monitor performance statistics and utilisation of results analytical framework and a global research network. GBD 2017 by research groups and ministries to track how improvements included more comprehensive detail of injury burden than any to injury measurement progress over time. Scientific dialogue other known efforts to date. GBD 2019 and future updates will and collaboration must be a major focus, and the GBD enter- continue to add detail and refine methods in the interest of prise is a good forum to support this kind of data sharing. providing injury burden estimates that are robust, accurate and

James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 i145 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research timely. Expanded injury data collection efforts will be a critical 34Department of Epidemiology, Arak University of Medical Sciences, Arak, Iran 35 component of future injury burden estimation. Physiotherapy Department, The University of Jordan, Amman, Jordan 36King Saud University, Riyadh, Saudi Arabia 37Clinical Practice Guidelines Unit, King Saud University, Riyadh, Saudi Arabia What is already known on the subject 38Alexandria Center for Evidence-­Based Clinical Practice Guidelines, Alexandria University, Alexandria, Egypt 39Carol Davila University of Medicine and Pharmacy, Bucharest, Romania ► 40 ► Global Burden of Disease (GBD) 2017 provided an extensive Department of Epidemiology and Biostatistics, Health Promotion Research Center, peer-­reviewed assessment of death and disability. Zahedan, Iran ►► GBD 2017 methods have been reviewed and updated 41Department of Health Policy and Administration, University of the Philippines iteratively as new methods and data become available. Manila, Manila, Philippines 42Department of Applied Social Sciences, Hong Kong Polytechnic University, Hong ►► Measuring injury burden in GBD 2017 is complex due to Kong, China differences in measuring cause of injury versus nature of 43Department of Parasitology, Mazandaran University of Medical Sciences, Sari, Iran injury and the temporal difference between them. 44Department of Microbiology and Immunology, Iranshahr University of Medical Sciences, Iranshahr, Iran 45Department of Sociology and Social Work, Kwame Nkrumah University of Science What this study adds and Technology, Kumasi, Ghana 46Center for International Health, Ludwig Maximilians University, Munich, Germany 47Social Determinants of Health Research Center, Birjand University of Medical ►► This capstone study details key estimation methods that are Sciences, Birjand, Iran used for measuring the global burden of injuries as described 48Department of Health Promotion and Education, Tehran University of Medical in related publications in this journal. Sciences, Tehran, Iran 49School of Health Sciences, Birmingham City University, Birmingham, UK ► 50 ► More detailed methods descriptions and model performance Regional Centre for the Analysis of Data on Occupational and Work-­related Injuries metrics from GBD 2017 are provided in this study than in and Diseases, Local Health Unit Tuscany Centre, Florence, Italy related studies. 51School of Science and Health, Western Sydney University, Sydney, New South ► Wales, Australia ► This study also includes suggested future directions for 52 improving injury burden research. Oral Health Services, Sydney Local Health District, Sydney, New South Wales, Australia 53Plastic Surgery Department, University of Texas, Houston, TX, USA 54The Judith Lumley Centre, La Trobe University, Melbourne, Victoria, Australia Author affiliations 55 General Office for Research and Technological Transfer, Peruvian National Institute 1Institute for Health Metrics and Evaluation, University of Washington, Seattle, WA, of Health, Lima, Peru USA 56 2 School of Public Health, Curtin University, Perth, Western Australia, Australia Department of Neurology, Cairo University, Cairo, Egypt 57Department of Health Policy Planning and Management, University of Health and 3Neuroscience Research Center, Isfahan University of Medical Sciences, Isfahan, Iran 4 Allied Sciences, Ho, Ghana Department of Public Health, Ministry of Health, Riyadh, Saudi Arabia 58 5 Department of Environmental Health Engineering, Hamadan University of Medical Department of Orthopaedic Surgery, University of Southern California, Los Angeles, Sciences, Hamadan, Iran CA, USA 59Public Health Risk Sciences Division, Public Health Agency of Canada, Toronto, 6Biostatistics and Health Informatics, Madda Walabu University, Bale Robe, Ethiopia

7 Ontario, Canada http://injuryprevention.bmj.com/ Radiotherapy Center, Addis Ababa University, Addis Ababa, Ethiopia 60 8 Department of Nutritional Sciences, University of Toronto, Toronto, Ontario, Canada Department of Public Health, Debre Berhan University, Debre Berhan, Ethiopia 61Department of Forensic Science, Government Institute of Forensic Science, Nagpur, 9Cardiovascular Medicine Department, Ain Shams University, Abbasia, Egypt 10 India Department of Medicine, University College Hospital, Ibadan, Nigeria 62 11 Biochemistry Unit, Universiti Sultan Zainal Abidin, Kuala Terengganu, Malaysia Sport Science Department, University of Extremadura, Badajoz, Spain 63School of Health Sciences, Univeristi Sultan Zainal Abidin, Kuala Terengganu, 12Social Behavioral Research Branch, National Institute of Health, Bethesda, MD, USA 13 Malaysia Cancer Prevention and Control, Georgetown University, Washington, DC, USA 64 14 Institute of Health Management Research, Indian Institute of Health Management School of Medicine, Center for Politics, Population and Health Research, National Research University, Jaipur, India Autonomous University of Mexico, Mexico City, Mexico 65 15 Department of Epidemiology, Johns Hopkins University, Baltimore, MD, USA Department of Epidemiology and Health Statistics, Southeast University Nanjing, 66Health Policy and Management Department, Tehran University of Medical Sciences, Nanjing, China Tehran, Iran 16Microbiology Department, Hazara University, Mansehra, Pakistan 67 17 Department of Demography, University of Groningen, Groningen, Netherlands Department of Epidemiology, Jimma University, Jimma, Ethiopia 68Population Research Centre, Institute for Social and Economic Change, Bengaluru, 18James P Grant School of Public Health, BRAC University, Dhaka, Bangladesh India on September 27, 2021 by guest. Protected copyright. 19 Health Systems and Population Studies Division, International Centre for Diarrhoeal 69Department of Hypertension, Medical University of Lodz, Lodz, Poland Disease Research, Dhaka, Bangladesh 70Polish Mothers’ Memorial Hospital Research Institute, Lodz, Poland 20 Higher National School of Veterinary Medicine, Algiers, Algeria 71School of Health Sciences, Walden University, Minneapolis, MN, USA 21 Evidence Based Practice Center, Mayo Clinic Foundation for Medical Education and 72Department of Noncommunicable Diseases, Bangladesh University of Health Research, Rochester, MN, USA Sciences (BUHS), Dhaka, Bangladesh 22 Department of Computer Sciences, Imam Abdulrehman Bin Faisal University, 73Department of Research, Public Health Perspective Nepal, Pokhara-­Lekhnath Dammam, Saudi Arabia Metropolitan City, Nepal 23 Department of Pharmacy, Adigrat University, Adigrat, Ethiopia 74School of Psychology, University of Auckland, Auckland, New Zealand 24 Medicine and Health Science, Arba Minch University, Arba Minch, Ethiopia 75Heidelberg Institute of Global Health (HIGH), Heidelberg University, Heidelberg, 25 Midwifery Department, Arba Minch University, Injbara, Ethiopia Germany 26 Department of Population Health Research, King Abdullah International Medical 76T.H. Chan School of Public Health, Harvard University, Boston, MA, USA Research Center, Riyadh, Saudi Arabia 77Occupational Health Department, Kermanshah University of Medical Sciences, 27Medical Technical Institute, Erbil Polytechnic University, Erbil, Iraq Kermanshah, Iran 28Department of Information Systems, College of Economics and Political Science, 78Health Human Resources Research Center, Shiraz University of Medical Sciences, Sultan Qaboos University, Muscat, Oman Shiraz, Iran 29Department of Health Care Management and Economics, Urmia University of 79Department of Psychiatry, Charles R. Drew University of Medicine and Science, Los Medical Science, Urmia, Iran Angeles, CA, USA 30Health Management and Economics Research Center, Iran University of Medical 80Department of Psychiatry and Biobehavioral Sciences, David Geffen School of Sciences, Tehran, Iran Medicine, University of California Los Angeles, Los Angeles, CA, USA 31Health Economics Department, Iran University of Medical Sciences, Tehran, Iran 81Department of Community Medicine, Gandhi Medical College Bhopal, Bhopal, India 32Department of Health Policy and Management, Kuwait University, Safat, Kuwait 82Jazan University, Jazan, Saudi Arabia 33International Centre for Casemix and Clinical Coding, National University of 83Social Determinants of Health Research Center, Lorestan University of Medical Malaysia, Bandar Tun Razak, Malaysia Sciences, Khorramabad, Iran i146 James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research

84Psychiatry Department, Bahir Dar University, Bhair Dar, Ethiopia 133Department of Epidemiology and Biostatistics, University of South Carolina, 85Nuffield Department of Population Health, University of Oxford, Oxford, UK Columbia, SC, USA 86Department of Internal Medicine, University of São Paulo, São Paulo, Brazil 134Department of Pulmonary Medicine, Christian Medical College and Hospital 87Department of Nutrition and Dietetics, Mekelle University, Mekelle, Ethiopia (CMC), Vellore, India 88Department of Internal Medicine, United Arab Emirates University, Al Ain, United 135Faculty of Biology, Hanoi National University of Education, Hanoi, Vietnam Arab Emirates 136School of Public Health and Preventive Medicine, Monash University, Melbourne, 89Social and Clinical Pharmacy, Charles University, Hradec Kralova, Czech Republic Victoria, Australia 90Department of Community Medicine, All India Institute of Medical Sciences, 137Centro Hospitalar Universitário do Porto - Serviço de Oftalmologia, University of Nagpur, India Porto, Porto, Portugal 91Department of Community Medicine, Datta Meghe Institute of Medical Sciences, 138Department of Environmental Health, Imam Abdulrahman Bin Faisal University, Wardha, India Dammam, Saudi Arabia 139 92Internal Medicine Department, University of Massachusetts Medical School, Toxoplasmosis Research Center, Mazandaran University of Medical Sciences, Sari, Springfield, MA, USA Iran 140 93Department of Statistical and Computational Genomics, National Institute of Population and Development, Facultad Latinoamericana de Ciencias Sociales Biomedical Genomics, Kalyani, India Mexico, Mexico City, Mexico 141 94Department of Statistics, University of Calcutta, Kolkata, India Australian Institute for Suicide Research and Prevention, Griffith University, Mount 95Centre for Global Child Health, University of Toronto, Toronto, Ontario, Canada Gravatt, Queensland, Australia 142 96Centre of Excellence in Women and Child Health, Aga Khan University, Karachi, Department of Medical Laboratory Sciences, Bahir Dar University, Bahir Dar, Pakistan Ethiopia 143 97Institute of Soil and Environmental Sciences, University of Agriculture, Faisalabad, School of Pharmacy, Aksum University, Aksum, Ethiopia 144Addis Ababa University, Addis Ababa, Ethiopia Pakistan 145 98Social Determinants of Health Research Center, Babol University of Medical Department of Global Health and Infection, Brighton and Sussex Medical School, Brighton, UK Sciences, Babol, Iran 146 99Department of Internal Medicine, Manipal Academy of Higher Education, School of Public Health, Addis Ababa University, Addis Ababa, Ethiopia 147Division of Cardiology, Atlanta Veterans Affairs Medical Center, Decatur, GA, USA Mangalore, India 148 100Department of Epidemiology and Psychosocial Reseach, Ramón de la Fuente Department of Epidemiology, Shiraz University of Medical Sciences, Shiraz, Iran 149Faculty of Pharmacy, University of Porto, Porto, Portugal Muñiz National Institute of Psychiatry, Mexico City, Mexico 150 101 Tehran University of Medical Sciences, Tehran, Iran Centre for Adolescent Health, Murdoch Childrens Research Institute, Melbourne, 151 Victoria, Australia Center of Excellence in Public Health Nutrition, Nguyen Tat Thanh University, Ho 102 Chi Minh City, Vietnam School of Population and Global Health, University of Melbourne, Melbourne, 152 Victoria, Australia School of Health and Biomedical Sciences, Royal Melbourne Institute of 103 Technology University, Bundoora, Victoria, Australia Department of Clinical and Experimental Medicine, University of Catania, Catania, 153 Italy Sydney School of Public Health, University of Sydney, Sydney, New South Wales, 104Transport and Road Safety (TARS) Research Department, University of New South Australia 154Faculty of Medicine, University of Belgrade, Belgrade, Serbia Wales, Sydney, New South Wales, Australia 155 105Division of Hematology and Oncology, Georgetown University, Washington DC, Public Health Department, Hawassa University, Hawassa, Ethiopia 156Curtin University, Perth, Western Australia, Australia USA 157 106 Department of Global Health and Social Medicine, Harvard University, Boston, Department of Epidemiology and Evidence Based Medicine, I.M. Sechenov First MA, USA Moscow State Medical University, Moscow, Russia 158 107 Department of Social Services, Tufts Medical Center, Boston, MA, USA Department of Health Sciences, University of Leicester, Leicester, UK 159

Department of Statistics, Debre Markos University, Debre Markos, Ethiopia http://injuryprevention.bmj.com/ 108Research Department, Golden Community, Kathmandu, Nepal 160Department of Public Health Sciences, Karolinska Institutet, Stockholm, Sweden 109Centre for Population Health Sciences, Nanyang Technological University, 161World Health Programme, Université du Québec en Abitibi-­Témiscamingue, Rouyn-­ Singapore, Singapore 110 Noranda, Quebec, Canada Global eHealth Unit, Imperial College London, London, UK 162 111 Department of Pathology, Stavanger University Hospital, Stavanger, Norway Department of Population and Health, Metropolitan Autonomous University, 163 Norwegian Institute of Public Health, Oslo, Norway Mexico City, Mexico 164 112 Department of Clinical Pathology, Mansoura University, Mansoura, Egypt Research Unit on Applied Molecular Biosciences (UCIBIO), University of Porto, 165Multiple Sclerosis Research Center, Tehran University of Medical Sciences, Tehran, Porto, Portugal 113 Iran Department of Psychiatry, University of São Paulo, São Paulo, Brazil 166 114 Epidemiology and Population Health, York University, Vancouver, British Columbia, Colombian National Health Observatory, National Institute of Health, Bogota, Canada Colombia 167 115 Faculty of Health Sciences, Simon Fraser University, Burnaby, British Columbia, Epidemiology and Public Health Evaluation Group, National University of Canada Colombia, Bogota, Colombia 168 116 Biology Department, Salahaddin University-­Erbil, Erbil, Iraq Primary Care Services Area, Central Health Directorate, Region Friuli Venezia 169 Department of Biology and Biotechnology “Lazzaro Spallanzani", University of on September 27, 2021 by guest. Protected copyright. Giulia, Trieste, Italy 117 Pavia, Pavia, Italy Department of Medicine (DAME), University of Udine, Udine, Italy 170 118 Department of Psychology, Federal University of Sergipe, Sao Cristovao, Brazil National School of Public Health, Carlos III Health Institute, Madrid, Spain 171 119 Non-­communicable Diseases Research Center, Tehran University of Medical Clinical Epidemiology Program, Ottawa Hospital Research Institute, Ottawa, Sciences, Tehran, Iran Ontario, Canada 172 120 Department of Neurobiology, Karolinska Institutet, Stockholm, Sweden Mary MacKillop Institute for Health Research, Australian Catholic University, 173Division of Neurology, University of Ottawa, Ottawa, Ontario, Canada Melbourne, Victoria, Australia 174REQUIMTE/LAQV, University of Porto, Porto, Portugal 121School of Public Health, University of Hong Kong, Hong Kong, China 175 122 Research Centre on Public Health (CESP), University of Milan Bicocca, Monza, Italy Institute of Applied Health Research, University of Birmingham, Birmingham, UK 176Department of Population Medicine and Health Services Research, Bielefeld 123 Swedish Neuroscience Institute, Swedish Brain and Spine Specialists, Seattle, WA, University, Bielefeld, Germany USA 177Department of Child Dental Health, Obafemi Awolowo University, Ile-­Ife, Nigeria 124 Department of Medicine, University of Toronto, Toronto, Ontario, Canada 178Timiryazev Institute of Plant Physiology, Russian Academy of Sciences, Moscow, 125 Department of Public Health, Texila American University, Georgetown, Guyana Russia 126 2nd Department of Ophthalmology, University of Athens, Haidari, Greece 179Abadan School of Medical Sciences, Abadan University of Medical Sciences, 127 Ophthalmology Independent Consultant, Athens, Greece Abadan, Iran 128 Pediatrics Department, Harvard University, Boston, MA, USA 180Department of Family Medicine and Primary Care, University of the Witwatersrand, 129 Neonatology Department, Beth Israel Deaconess Medical Center, Boston, MA, USA Johannesburg, South Africa 130 Department of Surgery, Division of Plastic and Reconstructive Surgery, University 181College of Public Health, Medical and Veterinary Science, James Cook University, of Washington, Seattle, WA, USA Douglas, Queensland, Australia 131Department of Biochemistry and Biomedical Science, Seoul National University 182Royal Life Saving Society, Sydney, New South Wales, Australia Hospital, Seoul, South Korea 183Department of Dermatology, Kobe University, Kobe, Japan 132Maternal and Child Health Division, International Centre for Diarrhoeal Disease 184Gene Expression & Regulation Program, The Wistar Institute, Philadelphia, PA, USA Research, Dhaka, Bangladesh 185Public Health Department, Madda Walabu University, Bale Robe, Ethiopia

James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 i147 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research

186Department of Nursing and Midwifery, Addis Ababa University, Addis Ababa, 238Department of Legal Medicine and Bioethics, Carol Davila University of Medicine Ethiopia and Pharmacy, Bucharest, Romania 187Pharmacy Department, Mekelle University, Mekelle, Ethiopia 239Clinical Legal Medicine Department, National Institute of Legal Medicine Mina 188Department of Nursing, Aksum University, Aksum, Ethiopia Minovici, Bucharest, Romania 189Department of Nursing, Mekelle University, Mekelle, Ethiopia 240Department of Epidemiology and Health Statistics, Central South University, 190Public Health, Haramaya University, Harar, Ethiopia Changsha, China 191Bahir Dar University, Bahir Dar, Ethiopia 241Department of Health Promotion and Education, University of Ibadan, Ibadan, 192Haramaya University, Dire Dawa, Ethiopia Nigeria 193Department of Pharmacy, Wollo University, Dessie, Ethiopia 242Department of Community Medicine, University of Ibadan, Ibadan, Nigeria 194Department of Nursing, Arba Minch University, Arba Minch, Ethiopia 243Department of Family Medicine, Bangalore Baptist Hospital, Bangalore, India 195Department of Medical Surgery, Tabriz University of Medical Sciences, Tabriz, Iran 244Research Institute for Endocrine Sciences, Shahid Beheshti University of Medical 196Occupational Health Department, Arak University of Medical Sciences, Arak, Iran Sciences, Tehran, Iran 197Department of Nursing and Midwifery, Kurdistan University of Medical Sciences, 245School of Psychology and Public Health, La Trobe University, Bundoora, Sanandaj, Iran Melbourne, Victoria, Australia 198Science and Research Branch, Islamic Azad University, Tehran, Iran 246Institute for Physical Activity and Nutrition, Deakin University, Burwood, Victoria, 199Young Researchers and Elite Club, Islamic Azad University, Rasht, Iran Australia 200Faculty of Allied Health Sciences, The University of Lahore, Lahore, Pakistan 247Sydney Medical School, University of Sydney, Sydney, New South Wales, Australia 201Chairman BOG, Afro-­Asian Institute, Lahore, Pakistan 248School of Public Health and Community Medicine, University of New South Wales, 202Adelaide Medical School, University of Adelaide, Adelaide, SA, Australia Sydney, New South Wales, Australia 203Nursing and Midwifery Department, Mazandaran University of Medical Sciences, 249Faculty of Medicine, Babol University of Medical Sciences, Babol, Iran Sari, Iran 250Department for Health Care and Public Health, Sechenov First Moscow State 204Center for Clinical and Epidemiological Research, University of São Paulo, Sao Medical University, Moscow, Russia Paulo, Brazil 251Social Development & Health Promotion Research Center, Kermanshah University 205Internal Medicine Department, University of São Paulo, Sao Paulo, Brazil of Medical Sciences, Kermanshah, Iran 206Department of Dermatology, Boston University, Boston, MA, USA 252Department of Surgery, Virginia Commonwealth University, Richmond, VA, USA 207Institute of Public Health, United Arab Emirates University, Al Ain, United Arab 253Institute of Medicine, University of Colombo, Colombo, Sri Lanka Emirates 254Faculty of Graduate Studies, University of Colombo, Colombo, Sri Lanka 208Instituto de Patologia Tropical e Saúde Pública, Federal University of Goias, 255Department of Community Medicine, Banaras Hindu University, Varanasi, India Goiânia, Brazil 256Health Promotion and Education, University of Ibadan, Ibadan, Nigeria 209Department of Epidemiology and Biostatistics, Zhengzhou University, Zhengzhou, 257Department of Ophthalmology, Heidelberg University, Mannheim, Germany China 258Beijing Ophthalmology & Visual Science Key Laboratory, Beijing Tongren Hospital, 210Non-­Communicable Diseases (NCD), World Health Organization (WHO), New Beijing, China Delhi, India 259Auckland University of Technology, Auckland, New Zealand 211Department of Public Health, Erasmus University Medical Center, Rotterdam, 260Community Medicine Department, Manipal Academy of Higher Education, Netherlands Mangalore, India 212Global and Community Mental Health Research Group, University of Macau, 261Department of Family Medicine and Public Health, University of Opole, Opole, Macao, China Poland 213Department of Family and Community Medicine, Arabian Gulf University, 262School of Health Sciences, Savitribai Phule Pune University, Pune, India Manama, Bahrain 263Institute of Family Medicine and Public Health, University of Tartu, Tartu, Estonia 214School of Health and Environmental Studies, Hamdan Bin Mohammed Smart 264Minimally Invasive Surgery Research Center, Iran University of Medical Sciences,

University, Dubai, United Arab Emirates Tehran, Iran http://injuryprevention.bmj.com/ 215Biomedical Research Networking Center for Mental Health Network (CiberSAM), 265Department of Medical Informatics, Tabriz University of Medical Sciences, Tabriz, Madrid, Spain Iran 216Research and Development Unit, San Juan de Dios Sanitary Park, Sant Boi de 266Social Determinants of Health Research Center, Research Institute for Prevention Llobregat, Spain of Non-­Communicable Diseases, Qazvin University of Medical Sciences, Qazvin, 217Department of Microbiology, Maragheh University of Medical Sciences, Maragheh, Iran Iran 267Health Services Management Department, Qazvin University of Medical Sciences, 218Department of Microbiology, Tehran University of Medical Sciences, Tehran, Iran Qazvin, Iran 219Centre for International Health and Section for Ethics and Health Economics, 268School of Public Health, Department of Health informatics and Health Innovation, University of Bergen, Bergen, Norway A.C.S. Medical College and Hospital, Mekelle, Ethiopia 220Gastrointestinal and Liver Disease Research Center, Guilan University of Medical 269Department of Forensic Medicine and Toxicology, All India Institute of Medical Sciences, Rasht, Iran Sciences, Jodhpur, India 221Guilan University of Medical Sciences, Rasht, Iran 270Department of Epidemiology, Hamadan University of Medical Sciences, Hamadan, 222School of Nursing and Midwifery, Tabriz University of Medical Sciences, Tabriz, Iran Iran 223 271

Independent Consultant, Tabriz, Iran Hematology-­Oncology and Stem Cell Transplantation Research Center, Tehran on September 27, 2021 by guest. Protected copyright. 224Department of Public Health, Mizan-­Tepi University, Tepi, Ethiopia University of Medical Sciences, Tehran, Iran 225Unit of Epidemiology and Social Medicine, University Hospital Antwerp, Wilrijk, 272Pars Advanced and Minimally Invasive Medical Manners Research Center, Iran Belgium University of Medical Sciences, Tehran, Iran 226Department of Clinical Sciences, Karolinska University Hospital, Stockholm, 273Department of Applied Physics, The John Paul II Catholic University of Lublin, Sweden Lublin Voivodeship, Poland 227Medical Biology Research Center, Kermanshah University of Medical Sciences, 274Department of Biology and Chemistry, Drohobych Ivan Franko State Pedagogical Kermanshah, Iran University, Drohobych, Ukraine 228Research Coordination, AC Environments Foundation, Cuernavaca, Mexico 275International Research Center of Excellence, Institute of Human Virology Nigeria, 229CISS, National Institute of Public Health, Cuernavaca. Mexico Abuja, Nigeria 230Department of Urban Planning and Design, University of Hong Kong, Hong Kong, 276Julius Centre for Health Sciences and Primary Care, Utrecht University, Utrecht, China Netherlands 231Center of Excellence in Behavioral Medicine, Nguyen Tat Thanh University, Ho Chi 277Open, Distance and eLearning Campus, University of Nairobi, Nairobi, Kenya Minh City, Vietnam 278Department of Dermatology, Wolaita Sodo University, Wolaita Sodo, Ethiopia 232Department of Pediatrics, Dell Medical School, University of Texas Austin, Austin, 279Department of Public Health, Jordan University of Science and Technology, Irbid, TX, USA Jordan 233Kasturba Medical College, Manipal Academy of Higher Education, Manipal, India 280Social Determinants of Health Research Center, Ahvaz Jundishapur University of 234Department of Pharmacology and Therapeutics, Dhaka Medical College, Dhaka, Medical Sciences, Ahvaz, Iran Bangladesh 281School of Food and Agricultural Sciences, University of Management and 235Department of Pharmacology, Bangladesh Industrial Gases Limited, Tangail, Technology, Lahore, Pakistan Bangladesh 282Department of Global Health, University of Washington, Seattle, WA, USA 236Department of Computer Engineering, Islamic Azad University, Tehran, Iran 283Department of Physiology, Baku State University, Baku, Azerbaijan 237Computer Science Department, University of Human Development, Sulaymaniyah, 284Epidemiology, Faculty of Public Health and Tropical Medicine, Jazan University, Iraq Jazan, Saudi Arabia i148 James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research

285Epidemiology and Biostatistics Department, Health Services Academy, Islamabad, 338Department of Primary Care and Public Health, Imperial College London, London, Pakistan UK 286Department of Population Studies, International Institute for Population Sciences, 339Health Education and Research Department, SDM College of Medical Sciences & Mumbai, India Hospital, Dharwad, India 287Department of Health Research, Indian Council of Medical Research, New Delhi, 340Health University, Rajiv Gandhi University of Health Sciences, Bangalore, India India 341Department of Maternal and Child Nursing and Public Health, Federal University 288Centre for Ethics, Jawahar Lal Nehru University, New Delhi, India of Minas Gerais, Belo Horizonte, Brazil 289Department of Psychiatry, Kermanshah University of Medical Sciences, 342Ophthalmology Department, Iran University of Medical Sciences, Tehran, Iran Kermanshah, Iran 343Ophthalmology Department, University of Manitoba, Winnipeg, Manitoba, Canada 290Nuffield Department of Surgical Sciences, Oxford University Global Surgery Group, 344Department of Surgery, University of Virginia, Charlottesville, VA, USA University of Oxford, Oxford, UK 345Surgery Department, Emergency University Hospital Bucharest, Bucharest, 291Research and Data Solutions, Synotech Consultant, Nairobi, Kenya Romania 292Department of Preventive Medicine, Korea University, Seoul, South Korea 346Psychiatry Department, National Institute of Mental Health and Neurosciences, 293School of Medicine, Xiamen University Malaysia, Sepang, Malaysia Bengaluru, India 294Department of Health Sciences, Northeastern University, Boston, MA, USA 347Department of Epidemiology and Biostatistics, Tehran University of Medical 295Department of Nursing and Health Promotion, Oslo Metropolitan University, Oslo, Sciences, Tehran, Iran Norway 348Institute for Social Science Research, The University of Queensland, Brisbane, 296School of Health Sciences, Kristiania University College, Oslo, Norway Queensland, Australia 297Neurophysiology Research Center, Hamadan University of Medical Sciences, 349Department of Health Sciences, University of York, York, UK Hamadan, Iran 350Department of Midwifery-­Reproductive Health, Hamadan University of Medical 298Brain Engineering Research Center, Institute for Research in Fundamental Sciences, Hamadan, Iran Sciences, Tehran, Iran 351Research Department, The George Institute for Global Health, New Delhi, India 299Public Health Dentistry Department, Krishna Institute of Medical Sciences Deemed 352School of Medicine, University of New South Wales, Sydney, New South Wales, to be University, Karad, India Australia 300Environmental Health Engineering, Arak University of Medical Sciences, Arak, Iran 353Neurology Department, Janakpuri Super Specialty Hospital Society, New Delhi, 301CIBERSAM, San Juan de Dios Sanitary Park, Sant Boi de Llobregat, Spain India 302Catalan Institution for Research and Advanced Studies (ICREA), Barcelona, Spain 354Department of Neurology, Govind Ballabh Institute of Medical Education and 303Department of Zoology, University of Oxford, Oxford, UK Research, New Delhi, India 304Harvard Medical School, Harvard University, Boston, MA, USA 355Division of Epidemiology and Prevention, Institute of Human Virology, University 305Department of Anthropology, Panjab University, Chandigarh, India of Maryland, Baltimore, MD, USA 306Department of Demography, University of Montreal, Montreal, Quebec, Canada 356Peru Country Office, United Nations Population Fund (UNFPA), Lima, Peru 307Department of Social and Preventive Medicine, University of Montreal, Montreal, 357Forensic Medicine Division, Imam Abdulrahman Bin Faisal University, Dammam, Quebec, Canada Saudi Arabia 308Department of Public Health, Yuksek Ihtisas University, Ankara, Turkey 358Department of Epidemiology and Biostatistics, Haramaya University, Harar, 309Department of Public Health, Hacettepe University, Ankara, Turkey Ethiopia 310Department of Family and Community Health, University of Health and Allied 359Breast Surgery Unit, Helsinki University Hospital, Helsinki, Finland Sciences, Ho, Ghana 360University of Helsinki, Helsinki, Finland 311Department of Psychology and Health Promotion, University of KwaZulu-­Natal, 361Neurocenter, Helsinki University Hospital, Helsinki, Finland Durban, South Africa 362School of Health Sciences, University of Melbourne, Parkville, Victoria, Australia 312Community Medicine Department, Kasturba Medical College, Manipal Academy of 363Statistics Department, Debre Markos University, Debre Markos, Ethiopia Higher Education, Mangalore, India 364Clinical Microbiology and Parasitology Unit, Zora Profozic Polyclinic, Zagreb, http://injuryprevention.bmj.com/ 313Department of Psychiatry, University of Nairobi, Nairobi, Kenya Croatia 314Division of Psychology and Language Sciences, University College London, 365University Centre Varazdin, University North, Varazdin, Croatia London, UK 366Center for Innovation in Medical Education, Pomeranian Medical University, 315Department of Medicine Brigham and Women’s Hospital, Harvard University, Szczecin, Poland Boston, MA, USA 367Pomeranian Medical University, Szczecin, Poland 316Orthopaedics Department, Base Hospital Lucknow Cantt, Lucknow, India 368Department of Propedeutics of Internal Diseases & Arterial Hypertension, 317Mechanical and Industrial Engineering, Indian Institute of Technology, Roorkee, Pomeranian Medical University, Szczecin, Poland India 369Pacific Institute for Research & Evaluation, Calverton, MD, USA 318Department of Community and Family Medicine, University of Baghdad, Baghdad, 370Achutha Menon Centre for Health Science Studies, Sree Chitra Tirunal Institute for Iraq Medical Sciences and Technology, Trivandrum, India 319HelpMeSee, New York, NY, USA 371Global Institute of Public Health (GIPH), Ananthapuri Hospitals and Research 320International Relations, Mexican Institute of Ophthalmology, Queretaro, Mexico Centre, Trivandrum, India 321Department of Otorhinolaryngology (ENT), Father Muller Medical College, 372Department of Statistics and Econometrics, Bucharest University of Economic Mangalore, India Studies, Bucharest, Romania on September 27, 2021 by guest. Protected copyright. 322Department of Public Health, Maragheh University of Medical Sciences, Maragheh, 373President’s Office, National Institute of Statistics, Bucharest, Romania Iran 374Faculty of Internal Medicine, Kyrgyz State Medical Academy, Bishkek, Kyrgyzstan 323Institute of Clinical Physiology, National Research Council, Pisa, Italy 375Department of Atherosclerosis and Coronary Heart Disease, National Center of 324Clinical Medicine and Community Health, University of Milan, Milano, Italy Cardiology and Internal Disease, Bishkek, Kyrgyzstan 325College of Optometry, Nova Southeastern University, Fort Lauderdale, FL, USA 376Heidelberg Institute of Global Health (HIGH), Faculty of Medicine and University 326School of Pharmacy, Monash University, Bandar Sunway, Malaysia Hospital, Heidelberg University, Heidelberg, Germany 327School of Pharmacy, Taylor’s University Lakeside Campus, Subang Jaya, Malaysia 377Institute of Addiction Research (ISFF), Frankfurt University of Applied Sciences, 328Department of Systems, Populations and Leadership, University of Michigan, Ann Frankfurt, Germany Arbor, MI, USA 378Biotechnology Research Center, Tabriz University of Medical Sciences, Tabriz, Iran 329Department of Health Metrics Sciences, School of Medicine, University of 379Molecular Medicine Research Center, Tabriz University of Medical Sciences, Tabriz, Washington, Seattle, WA, USA Iran 330Department of Medicine, University of São Paulo, Sao Paulo, Brazil 380Health Equity Research Center, Tehran University of Medical Sciences, Tehran, Iran 331Health Data Research UK, Swansea University, Swansea, UK 381Internal Medicine Department, King Saud University, Riyadh, Saudi Arabia 332Center for Integration of Data and Health Knowledge, FIOCRUZ: Cidacs Center for 382Department of Information Technology, University of Human Development, Integration of Data and Health Knowledge, Salvador, Brazil Sulaymaniyah, Iraq 333Faculty of Epidemiology and Population Health, London School of Hygiene & 383Department of Epidemiology and Biostatistics, Shahrekord University of Medical Tropical Medicine, England SciencesShahrekord, Iran 334Pathology Department, College of Medicine, Imam Abdulrahman Bin Faisal 384Department of Nursing, Shahroud University of Medical Sciences, Shahroud, Iran University, Dammam, Saudi Arabia 385Health Systems and Policy Research Unit, Ahmadu Bello University, Zaria, Nigeria 335Ophthalmology Department, Aswan Faculty of Medicine, Aswan, Egypt 386Department of Public Health, Samara University, Samara, Ethiopia 336Institute of Medicine, Tribhuvan University, Kathmandu, Nepal 387Iran National Institute of Health Research, Tehran University of Medical Sciences, 337Department of Public Health, Trnava University, Trnava, Slovakia Tehran, Iran

James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 i149 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research

388Pediatric Neurorehabilitation Research Center, University of Social Welfare and 435Department of Psychiatry and Behavioural Neurosciences, McMaster University, Rehabilitation Sciences, Tehran, Iran Hamilton, Ontario, Canada 389Faculty of Life Sciences and Medicine, King’s College London, London, UK 436Department of Psychiatry, University of Lagos, Lagos, Nigeria 390Clinical Epidemiology and Public Health Research Unit, Burlo Garofolo Institute for 437Department of Pathology and Molecular Medicine, McMaster University, Hamilton, Maternal and Child Health, Trieste, Italy Ontario, Canada 391Department of Public Health Medicine, University of KwaZulu-­Natal, Durban, 438Diplomacy and Public Relations Department, University of Human Development, South Africa Sulaimaniyah, Iraq 392Research Center for Environmental Determinants of Health, Kermanshah 439Department of Pharmacology and Therapeutics, University of Nigeria Nsukka, University of Medical Sciences, Kermanshah, Iran Enugu, Nigeria 393Kermanshah University of Medical Sciences, Kermanshah, Iran 440Applied Research Division, Public Health Agency of Canada, Ottawa, Ontario, 394Social Determinants of Health Research Center, Kurdistan University of Medical Canada 441 Sciences, Sanandaj, Iran School of Psychology, University of Ottawa, Ottawa, Ontario, Canada 442 395Department of Epidemiology and Biostatistics, Kurdistan University of Medical Department of Global Health Nursing, St. Luke’s International University, Chuo-­ku, Sciences, Sanandaj, Iran Japan 443 396Preventive Medicine and Public Health Research Center, Iran University of Medical Academic Department, Unium Ltd, Moscow, Russia 444 Sciences, Tehran, Iran Department of Project Management, National Research University Higher School 397International Laboratory for Air Quality and Health, Queensland University of of Economics, Moscow, Russia 445 Technology, Brisbane, Queensland, Australia Department of Respiratory Medicine, Jagadguru Sri Shivarathreeswara Academy of 398Gorgas Memorial Institute for Health Studies, Panama City, Panama Health Education and Research, Mysore, India 446 399Department of Psychiatry, Badhir Dar University, Ethiopia Department of Forensic Medicine, Manipal Academy of Higher Education, Manipal, 400 India Department of Epidemiology and Biostatistics, University of Gondar, Gondar, 447 Ethiopia Department of Medicine, Ottawa Hospital Research Institute, Ottawa Hospital, 401 Ottawa, Ontario, Canada School of Medical Sciences, Science University of Malaysia, Kubang Kerian, 448 Malaysia Parasitology and Mycology Department, Shiraz University of Medical Sciences, 402 Shiraz, Iran Department of Pediatric Medicine, Nishtar Medical University, Multan, Pakistan 449 403 Augenpraxis Jonas, Heidelberg University, Heidelberg, Germany Department of Pediatrics & Pediatric Pulmonology, Institute of Mother & Child 450 Care, Multan, Pakistan Department of Medical Humanities and Social Medicine, Kosin University, Busan, 404Clinical Research Development Center, Kermanshah University of Medical South Korea 451Research and Evaluation Department, Population Council, New Delhi, India Sciences, Kermanshah, Iran 452 405Research and Analytics, Initiative for Financing Health and Human Development, Indian Institute of Health Management Research University, Jaipur, India 453Department of Pediatircs, RD Gardi Medical College, Ujjain, India Chennai, India 454 406 Public Health Sciences, Karolinska Institutet, Stockholm, Sweden Research and Analytics, Bioinsilico Technologies, Chennai, India 455 407Department of Epidemiology, University of Alabama at Birmingham, Birmingham, Regional Medical Research Centre, Indian Council of Medical Research, Bhubaneswar, India AL, USA 456 408 Department of Midwifery, Wolaita Sodo University, Wolaita Sodo, Ethiopia Laboratory of Public Health Indicators Analysis and Health Digitalization, Moscow 457 Institute of Physics and Technology, Dolgoprudny, Russia School of Public Health and Community Medicine, Faculty of Medicine, University 409Experimental Surgery and Oncology Laboratory, Kursk State Medical University of of New South Wales, Sydney, New South Wales, Australia 458Center for Research and Innovation, Ateneo De Manila University, Pasig City, the Ministry of Health of the Russian Federation, Kursk, Russia 410 Philippines Department of Epidemiology & Biostatistics, Kermanshah University of Medical 459Department of Orthopedics, Yenepoya Medical College, Mangalore, India Sciences, Kermanshah, Iran 460

Department of Psychiatry, Department of Epidemiology, Columbia University, New http://injuryprevention.bmj.com/ 411Suraj Eye Institute, Nagpur, India York, NY, USA 412Hospital of the Federal University of Minas Gerais, Federal University of Minas 461Shanghai Mental Health Center, Shanghai Jiao Tong University, Shanghai, China Gerais, Belo Horizonte, Brazil 462Department of Epidemiology and Evidence-­Based Medicine, Sechenon University, 413Mental Health Research Center, IUMS, Tehran, Iran 414 Moscow, Russia Preventive Medicine and Public Health Research Center, IUMS, Tehran, Iran 463 School of Population and Public Health, University of British Columbia, Vancouver, 415Department of Forensic Medicine and Toxicology, Manipal Academy of Higher British Columbia, Canada Education, Manipal, India 464Digestive Diseases Research Institute, Tehran University of Medical Sciences, 416Department of Pediatrics, Arak University of Medical Sciences, Arak, Iran 417 Tehran, Iran Iranian Ministry of Health and Medical Education, Tehran, Iran 465 418 Department of Nephrology, Sanjay Gandhi Postgraduate Institute of Medical Cochrane South Africa, South African Medical Research Council, Cape Town, South Sciences, Lucknow, India Africa 466 419 Health Sciences Department, Muhammadiyah University of Surakarta, Sukoharjo, Department of General Surgery, Carol Davila University of Medicine and Indonesia Pharmacy, Bucharest, Romania 467 420 Department of Chemistry, Sharif University of Technology, Tehran, Iran Department of General Surgery, Emergency Hospital of Bucharest, Bucharest, 468 Biomedical Engineering Department, Amirkabir University of Technology, Tehran, on September 27, 2021 by guest. Protected copyright. Romania 421 Iran Department of Biological Sciences, University of Embu, Embu, Kenya 469 422 College of Medicine, University of Central Florida, Orlando, FL, USA Institute for Global Health Innovations, Duy Tan University, Hanoi, Vietnam 470 423 College of Graduate Health Sciences, A.T. Still University, Mesa, AZ, USA Project of ADB, National Institute of Nutrition, Hanoi, Vietnam 471 424 Department of Epidemiology & Biostatistics, Contech School of Public Health, Industrial Management Department, Hanoi University of Science and Technology, Lahore, Pakistan Hanoi, Vietnam 472 425 Department of Medicine, University of Alberta, Edmonton, Alberta, Canada Department of Pharmacology, Shahid Beheshti University of Medical Sciences, 473Department of Immunology, Mazandaran University of Medical Sciences, Sari, Iran Tehran, Iran 474 426 Molecular and Cell Biology Research Center, Mazandaran University of Medical Heidelberg University Hospital, Heidelberg, Germany Sciences, Sari, Iran 427 Public Health Department, Universitas Negeri Semarang, Kota Semarang, 475Thalassemia and Hemoglobinopathy Research Center, Ahvaz Jundishapur Indonesia University of Medical Sciences, Ahvaz, Iran 428 Graduate Institute of Biomedical Informatics, Taipei Medical University, Taipei City, 476Metabolomics and Genomics Research Center, Tehran University of Medical Taiwan Sciences, Tehran, Iran 429 School of Public Health and Family Medicine, University of Cape Town, Cape Town, 477Faculty of Medicine, Mazandaran University of Medical Sciences, Sari, Iran South Africa 478School of Nursing and Healthcare Professions, Federation University Australia, 430 Centre of Cardiovascular Research and Education in Therapeutics, Monash Berwick, Victoria, Australia University, Melbourne, Victoria, Australia 479School of Nursing and Midwifery, La Trobe University, Melbourne, Victoria, 431 Independent Consultant, Accra, Ghana Australia 432 UCIBIO, University of Porto, Porto, Portugal 480Faculty of Medicine, Birjand University of Medical Sciences, Birjand, Iran 433Reproductive Health Sciences, Department Obstetrics and Gynecology, University 481European Office for the Prevention and Control of Noncommunicable Diseases, of Ibadan, Ibadan, Nigeria World Health Organization (WHO), Moscow, Russia 434Department of Preventive Medicine, Kyung Hee University, Dongdaemun-­gu, 482Department of Oral Pathology, Srinivas Institute of Dental Sciences, Mangalore, South Korea India i150 James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research

483School of Behavioral Sciences and Mental Health, Tehran Institute of Psychiatry, 537Department of Medicine, University of Alabama at Birmingham, Birmingham, AL, Tehran, Iran USA 484Academic Public Health Department, Public Health England, London, UK 538Medicine Service, US Department of Veterans Affairs (VA), Birmingham, AL, USA 485School of Health, Medical and Applied Sciences, CQ University, Sydney, New South 539Department of Forensic Medicine, Kathmandu University, Dhulikhel, Nepal Wales, Australia 540Department of Epidemiology, School of Preventive Oncology, Patna, India 486Department of Computer Science, Metropolitan College, Boston University, 541Department of Epidemiology, Healis Sekhsaria Institute for Public Health, Mumbai, Boston, USA India 487Neurology Department, Sree Chitra Tirunal Institute for Medical Sciences and 542Medical Surgical Nursing Department, Urmia University of Medical Science, Urmia, Technology, Thiruvananthapuram, India Iran 488Brien Holden Vision Institute, Sydney, New South Wales, Australia 543Emergency Nursing Department, Semnan University of Medical Sciences, Semnan, 489Organization for the Prevention of Blindness, Paris, France Iran 544 490EPIUnit - Public Health Institute University Porto (ISPUP), University of Porto, Porto, Hospital Universitario de la Princesa, Autonomous University of Madrid, Madrid, Portugal Spain 545 491Surgery Department, University of Minnesota, Minneapolis, MN, USA Centro de Investigación Biomédica en Red Enfermedades Respiratorias (CIBERES), 492Surgery Department, University Teaching Hospital of Kigali, Kigali, Rwanda Madrid, Spain 546 493Research Directorate, Nihon Gakko University, Fernando de la Mora, Paraguay Department of Public Health, Arba Minch University, Arba Minch, Ethiopia 547 494Research Direction, Universidad Nacional de Caaguazú, Coronel Oviedo, Paraguay Hull York Medical School, University of Hull, Hull City, UK 548 495Department of Clinical Research, Federal University of Uberlândia, Uberlândia, Usher Institute of Population Health Sciences and Informatics, University of Brazil Edinburgh, Edinburgh, UK 549 496Golestan Research Center of Gastroenterology and Hepatology, Golestan Department of Psychology, Deakin University, Melbourne, Victoria, Australia 550Department of Community Medicine, Ahmadu Bello University, Zaria, Nigeria University of Medical Sciences, Gorgan, Iran 551 497College of Medicine, University of Sharjah, Sharjah, United Arab Emirates Department of Criminology, Law and Society, University of California Irvine, Irvine, 498Department of Health in Disasters and Emergencies, Shahid Beheshti University of CA, USA 552Department of Medicine, University of Valencia, Valencia, Spain Medical Sciences, Tehran, Iran 553 499Department of Neuroscience, Iran University of Medical Sciences, Tehran, Iran Carlos III Health Institute, Biomedical Research Networking Center for Mental 500Sina Trauma and Surgery Research Center, Tehran University of Medical Sciences, Health Network (CiberSAM), Madrid, Spain 554School of Social Work, University of Illinois, Urbana, IL, USA Tehran, Iran 555 501 Public Health, Arba Minch College of Health Sciences, Arba Minch, Ethiopia Nanobiotechnology Center, Soran University, Soran, Iraq 556 502 School of Public Health, University of Adelaide, Adelaide, SA, Australia Public Health and Community Medicine Department, Cairo University, Giza, Egypt 557 503 Department of Environmental Health, Wollo University, Dessie, Ethiopia Urology Department, Cairo University, Giza, Egypt 558 504Health and Disability Intelligence Group, Ministry of Health, Wellington, New Department of Community and Family Medicine, Iran University of Medical Sciences, Tehran, Iran Zealand 559 505 Department of Pharmacognosy, Mekelle University, Mekelle, Ethiopia Department of Entomology, Ain Shams University, Cairo, Egypt 560 506 Institute of Public Health, University of Gondar, Gondar, Ethiopia Department of Surgery, Marshall University, Huntington, WV, USA 561 507Department of Nutrition and Preventive Medicine, Case Western Reserve Department of Public Health, Adigrat University, Adigrat, Ethiopia 562Biology Department, Moscow State University, Moscow, Russia University, Cleveland, OH, USA 563 508Rheumatology Department, University Hospitals Bristol NHS Foundation Trust, Department of Nursing, Woldia University, Woldia, Ethiopia 564HIV/STI Surveillance Research Center, and WHO Collaborating Center for HIV Bristol, UK 509 Surveillance, Kerman University of Medical Sciences, Kerman, Iran Institute of Bone and Joint Research, University of Sydney, Syndey, New South 565Institute of Public Health, Krakow, Poland Wales, Australia 566

The Agency for Health Technology Assessment and Tariff System, Warszawa, http://injuryprevention.bmj.com/ 510Institute of Social Medicine, University of Belgrade, Belgrade, Serbia Poland 511Centre-­School of Public Health and Health Management, University of Belgrade, 567Department of Molecular Medicine and Pathology, University of Auckland, Belgrade, Serbia Auckland, New Zealand 512Health Economics, Bangladesh Institute of Development Studies (BIDS), Dhaka, 568Clinical Hematology and Toxicology, Military Medical University, Hanoi, Vietnam Bangladesh 569 513 Department of Health Economics, Hanoi Medical University, Hanoi, Vietnam Colorectal Research Center, Iran University of Medical Sciences, Tehran, Iran 570 514 Department of Psychiatry, Massachusetts General Hospital, Boston, MA, USA Surgery Department, Hamad Medical Corporation, Doha, Qatar 571 515 Mbarara University of Science and Technology, Mbarara, Uganda Faculty of Health & Social Sciences, Bournemouth University, Bournemouth, UK 572 516 Department of Medicine, University of Crete, Heraklion, Greece Department of Public Health Sciences, University of North Carolina at Charlotte, 573Lee Kong Chian School of Medicine, Nanyang Technological University, Singapore, Charlotte, NC, USA 517 Singapore Education Development Center, Faculty Member of Education Development 574Gomal Center of Biochemistry and Biotechnology, Gomal University, Dera Ismail Center, Ahvaz Jundishapur University of Medical Sciences, Ahvaz, Iran 518 Khan, Pakistan Department of Psychology, University of Alabama at Birmingham, Birmingham, 575TB Culture Laboratory, Mufti Mehmood Memorial Teaching Hospital, Dera Ismail AL, USA 519 Khan, Pakistan on September 27, 2021 by guest. Protected copyright. Department of Psychiatry, Stellenbosch University, Cape Town, South Africa 576 520 Amity Institute of Biotechnology, Amity University Rajasthan, Jaipur, India Emergency Department, Manian Medical Centre, Erode, India 577 521 Division of Health Sciences, University of Warwick, Coventry, UK Microbiology Service, National Institutes of Health, Bethesda, MD, USA 578 522 Argentine Society of Medicine, Buenos Aires, Argentina Center for Biomedical Information Technology, Shenzhen Institutes of Advanced 579Velez Sarsfield Hospital, Buenos Aires, Argentina Technology, Chinese Academy of Sciences, Shenzhen, China 580 523 UKK Institute, Tampere, Finland Department of Health Promotion and Education, Alborz University of Medical 581Psychosocial Injuries Research Center, Ilam University of Medical Sciences, Ilam, Sciences, Karaj, Iran Iran 524Health Policy Research Center, Shiraz University of Medical Sciences, Shiraz, Iran 582 525 Raffles Neuroscience Centre, Raffles Hospital, Singapore, Singapore Independent Consultant, Karachi, Pakistan 583Yong Loo Lin School of Medicine, National University of Singapore, Singapore, 526 School of Medicine, Alborz University of Medical Sciences, Karaj, Iran Singapore 527 Centre for Medical Informatics, University of Edinburgh, Edinburgh, UK 584Department of Medical and Surgical Sciences, University of Bologna, Bologna, Italy 528 Division of General Internal Medicine, Harvard University, Boston, MA, USA 585Occupational Health Unit, Sant’Orsola Malpighi Hospital, Bologna, Italy 529 National Institute of Infectious Diseases, Tokyo, Japan 586Department of Health Care Administration and Economics, National Research 530 College of Medicine, Yonsei University, Seodaemun-­gu, South Korea University Higher School of Economics, Moscow, Russia 531 Division of Cardiology, Emory University, Atlanta, GA, USA 587Foundation University Medical College, Foundation University, Islamabad, Pakistan 532 Finnish Institute of Occupational Health, Helsinki, Finland 588Department of Physical Therapy, Naresuan University, Meung District, Thailand 533 Department of Health Education & Promotion, Kermanshah University of Medical 589Department of Human Anatomy, Histology, and Embryology, Bahir Dar University, Sciences, Kermanshah, Iran Bahir Dar, Ethiopia 534 School of Health, University of Technology Sydney, Sydney, New South Wales, 590Department of Nursing, Wollo University, Dessie, Ethiopia Australia 591Department of Orthopaedics, Wenzhou Medical University, Wenzhou, China 535Department of Psychology, Reykjavik University, Reykjavik, Iceland 592Public Health Science Directorate, NHS Health Scotland, Glasgow, Scotland 536Department of Health and Behavior Studies, Columbia University, New York, NY, 593Medical Physics Department, Ahvaz Jundishapur University of Medical Sciences, USA Ahvaz, Iran

James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 i151 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research

594Department of Preventive Medicine, Northwestern University, Chicago, IL, USA Development of the Republic of Serbia through the Grant number OI175014; 595School of International Development and Global Studies, University of Ottawa, publication of results was not contingent upon Ministry’s censorship or approval. Ottawa, Ontario, Canada Sudha Jayaraman acknowledges support from: NIH R21: 1R21TW010439-­01A1 (PI); 596 Health Services Management Research Center, Kerman University of Medical Rotary Foundation Global Grant #GG1749568 (PI); NIH P20: 1P20CA210284-­01A1 Sciences, Kerman, Iran (Co-­PI) and DOD grant W81XWH-16-2-0040 (Co-­I), during the period of this study. 597Department of Health Management, Policy and Economics, Kerman University of Yun Jin Kim acknowledges support from a grant from the Research Management Medical Sciences, Kerman, Iran 598Division of Injury Prevention and Mental Health Improvement, National Center for Centre, Xiamen University Malaysia [grant number: XMUMRF/2018-­C2/ITCM/0001]. Chronic and Noncommunicable Disease Control, Chinese Center for Disease Control Kewal Krishan is supported by UGC Centre of Advanced Study (CAS II) awarded and Prevention, Beijing, China to the Department of Anthropology, Panjab University, Chandigarh, India. Mansai 599Centre for Suicide Research and Prevention, University of Hong Kong, Hong Kong, Kumar acknowledges support from FIC/ NIH K43 1K43MH114320-01. Amanda China Mason-­Jones acknowledges support by the University of York. Mariam Molokhia 600Department of Social Work and Social Administration, University of Hong Kong, is supported by the National Institute for Health Research Biomedical Research Hong Kong, China Center at Guy’s and St Thomas’s National Health Service Foundation Trust and King’s 601School of Allied Health Sciences, Addis Ababa University, Addis Ababa, Ethiopia 602 College London. Ilais Moreno Velasquez is supported by the Sistema Nacional de Department of Psychopharmacology, National Center of Neurology and Psychiatry, Investigación (SNI, Senacyt, Panama). Mukhammad David Naimzada acknowledges Tokyo, Japan support from Government of the Russian Federation (Agreement No – 075-02- 603Department of Sociology, Yonsei University, Seoul, South Korea 604 2019-967). Duduzile Ndwandwe acknowledges support from Cochrane South Department of Health Policy & Management, Jackson State University, Jackson, MS, USA Africa, South African Medical Research Council. Stanislav S. Otstavnov acknowledges 605School of Medicine, Tsinghua University, Beijing, China the support from the Government of the Russian Federation (Agreement No – 606Department of Environmental Health, Mazandaran University of Medical Sciences, 075-02-2019-967). Ashish Pathak acknowledges support from Indian Council of Sari, Iran Medical Research (ICMR), New Delhi, India (Grant number 2013-1253). Michael R 607Environmental Health, Academy of Medical Science, Sari, Iran Phillips acknowledges support in part by a grant from the National Natural Science 608 Department of Epidemiology and Biostatistics, Wuhan University, Wuhan, China Foundation of China (No.81761128031). Abdallah M. Samy received a fellowship 609 Global Health Institute, Wuhan University, Wuhan, China from the Egyptian Fulbright Mission Program. Milena Santric Milicevic acknowledges 610 School of Public Health and Management, Hubei University of Medicine, Shiyan, the support from the Ministry of Education, Science and Technological Development, China Republic of Serbia (Contract No. 175087). Seyedmojtaba Seyedmousavi was 611Social Determinants of Health Research Center, Ardabil University of Medical supported by the Intramural Research Program of the National Institutes of Science, Ardabil, Iran 612Department of Epidemiology, University Hospital of Setif, Setif, Algeria Health, Clinical Center, Department of Laboratory Medicine, Bethesda, MD, USA. 613Department of Medicine, School of Clinical Sciences at Monash Health, Monash Rafael Tabarés-­Seisdedos was supported in part by the national grant PI17/00719 University, Melbourne, Victoria, Australia from ISCIII-­FEDER. Sojib Bin Zaman acknowledges support from an "Australian 614Student Research Committee, Babol University of Medical Sciences, Babol, Iran Government Research Training Program (RTP) Scholarship.” 615 Department of Community Medicine, Ardabil University of Medical Science, Funding This study was funded by The Bill and Melinda Gates Foundation. SLJ Ardabil, Iran 616 conducts research for a grant on influenza and RSV which is funded in part by Sanofi Faculty of Medical Sciences, Department of Health Education, Tarbiat Modares Pasteur. University, Tehran, Iran 617Department of Preventive Medicine, Wuhan University, Wuhan, China Competing interests Dr. James reports grants from Sanofi Pasteur, outside the 618School of Public Health, Wuhan University of Science and Technology, Wuhan, submitted work. Dr. Driscoll reports grants from World Health Organisation, during China the conduct of the study. Dr Shariful Islam is funded by a Fellowship from National http://injuryprevention.bmj.com/ 619Hubei Province Key Laboratory of Occupational Hazard Identification and Control, Heart Foundation of Australia and Institute for Physical Activity and Nutrition, Wuhan University of Science and Technology, Wuhan, China Deakin University. Dr. Ivers reports grants from National Health and Medical 620Indian Institute of Public Health, Public Health Foundation of India, Gurugram, Research Council of Australia, during the conduct of the study. Dr. Jozwiak reports India personal fees from TEVA, personal fees from ALAB, personal fees from BOEHRINGER 621Public Health Foundation of India, Gurugram, India INGELHEIM, personal fees from SYNEXUS, non-­financial support from SERVIER, 622Department of Community Medicine, University of Peradeniya, Peradeniya, Sri non-­financial support from MICROLIFE, non-­financial support from MEDICOVER, Lanka outside the submitted work. Walter Mendoza is currently Program Analyst Population and Development at the Peru Country Office of the United Nations Population Fund-­UNFPA, which does not necessarily endorses this study. Dr. Rakovac reports Acknowledgements Seyed Aljunid acknowledges the Department of Health Policy grants from World Health Organization, during the conduct of the study. Dr. Sheikh and Management, Faculty of Public Health, Kuwait University and International reports grants from Health Data Research UK, outside the submitted work. Dr. Singh Centre for Casemix and Clinical Coding, Faculty of Medicine, National University reports personal fees from Crealta/Horizon, Medisys, Fidia, UBM LLC, Trio health, of Malaysia for the approval and support to participate in this research project. Medscape, WebMD, Clinical Care options, Clearview healthcare partners, Putnam Alaa Badawi acknowledges support by the Public Health Agency of Canada. Till associates, Spherix, Practice Point communications, the National Institutes of Health on September 27, 2021 by guest. Protected copyright. Bärnighausen acknowledges support by the Alexander von Humboldt Foundation and the American College of Rheumatology, personal fees from Speaker’s bureau of Simply Speaking. Dr. Singh owns stock options in Amarin pharmaceuticals and through the Alexander von Humboldt Professor award, funded by the German Viking pharmaceuticals. Dr. Singh serves on the steering committee of OMERACT, Federal Ministry of Education and Research. Traolach Brugha received support from an international organization that develops measures for clinical trials and receives NatCen Social Research (http://​natcen.​ac.​uk/) via NHS Digital and Department of arms-­length funding from 12 pharmaceutical companies. Dr. Singh serves on the Health and Social Care London, for the Adult Psychiatric Morbidity Survey (APMS) FDA Arthritis Advisory Committee. Dr. Singh is a member of the Veterans Affairs programme. Felix Carvalho received support from UID/MULTI/04378/2019 with Rheumatology Field Advisory Committee. Dr. Singh is the editor and the Director funding from FCT/MCTES through national funds. Vera M Costa acknowledges of the UAB Cochrane Musculoskeletal Group Satellite Center on Network Meta-­ support from grant (SFRH/BHD/110001/2015), received by Portuguese national analysis, outside the submitted work. Dr. Stein reports personal fees from Lundbeck funds through Fundação para a Ciência e Tecnologia (FCT), IP, under the Norma and Sun, outside the submitted work. Transitória DL57/2016/CP1334/CT0006. Kebede Deribe is supported by a grant from Patient consent for publication Not required. the Wellcome Trust [grant number 201900] as part of his International Intermediate Provenance and peer review Not commissioned; externally peer reviewed. Fellowship. Tim Driscoll acknowledges that work on occupational risk factors Data availability statement Availability of input data varies by source. Select was partially supported by funds from the World Health Organization. Eduarda data are available in a public, open-­access repository. Select data are available on Fernandes acknowledges support from UID/QUI/50006/2019 with funding from FCT/ reasonable request. Select data may be obtained from a third party and are not MCTES through national funds. Yuming Guo acknowledges support from Career publicly available. All results from the study are included in the article or uploaded as Development Fellowships of the Australian National Health and Medical Research supplementary information or are available online. Council (numbers APP1107107 and APP1163693). Sheikh Mohammed Shariful Open access This is an open access article distributed in accordance with the Islam is funded by a Fellowship from National Heart Foundation of Australia and Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits Institute for Physical Activity and Nutrition, Deakin University. Mihajlo Jakovljevic others to copy, redistribute, remix, transform and build upon this work for any acknowledges support by the Ministry of Education Science and Technological purpose, provided the original work is properly cited, a link to the licence is given, i152 James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 Inj Prev: first published as 10.1136/injuryprev-2019-043531 on 24 August 2020. Downloaded from Original research and indication of whether changes were made. See: https://​creativecommons.​org/​ 12 Foreman KJ, Lozano R, Lopez AD, et al. Modeling causes of death: an integrated licenses/by/​ ​4.0/.​ approach using CODEm. Popul Health Metr 2012;10:1. 13 Murray CJL, Lozano R, Flaxman AD, et al. Using verbal autopsy to measure causes of death: the comparative performance of existing methods. BMC Med 2014;12:5. 14 China Zhuhai Study. 2006-2007 - China CDC | GHDx. Available: http://ghdx.​ ​ References healthdata.org/​ ​record/china-​ ​zhuhai-study-​ ​2006-2007-​ ​china-cdc​ [Accessed 15 May 1 Bank W. World development report 1993: investing in health. Oxford University Press, 2018]. 1993. 15 GHDx. Functional outcome at 2.5, 5, 9, and 24 months after injury in the Netherlands. 2 Murray CJL, Callender CSKH, Kulikoff XR, et al. Population and fertility by age and Available: http://​ghdx.healthdata.​ ​org/record/​ ​functional-outcome-​ ​25-5-​ ​9-and-​ 24-​ ​ sex for 195 countries and territories, 1950–2017: a systematic analysis for the global months-​after-​injury-​netherlands [Accessed 15 May 2018]. burden of disease study 2017. The Lancet 2018;392:1995–2051. 16 GHDx. Health-­Related quality of life after burns: a prospective multicentre cohort 3 James SL, Abate D, Abate KH, et al. Global, regional, and national incidence, study with 18 months follow-up.­ Available: http://​ghdx.healthdata.​ ​org/record/​ health-​ ​ prevalence, and years lived with disability for 354 diseases and injuries for 195 related-quality-​ ​life-after-​ ​burns-prospective-​ ​multicentre-cohort-​ ​study-18-​ ​months-​ countries and territories, 1990–2017: a systematic analysis for the global burden of follow [Accessed 15 May 2018]. disease study 2017. The Lancet 2018;392:1789–858. 17 GHDx. Netherlands injury surveillance system 2007. Available: http://ghdx.​ ​healthdata.​ 4 Kyu HH, Abate D, Abate KH, et al. Global, regional, and national disability-­adjusted org/record/​ ​netherlands-injury-​ ​surveillance-system-​ ​2007 life-­years (DALYs) for 359 diseases and injuries and healthy life expectancy (HALE) for [Accessed 15 May 2018]. 195 countries and territories, 1990–2017: a systematic analysis for the global burden 18 GHDx. Netherlands injury surveillance system 2010. Available: http://ghdx.​ ​healthdata.​ of disease study 2017. The Lancet 2018;392:1859–922. org/record/​ ​netherlands-injury-​ ​surveillance-system-​ ​2010 5 Stanaway JD, Afshin A, Gakidou E, et al. Global, regional, and national comparative [Accessed 14 May 2018]. risk assessment of 84 behavioural, environmental and occupational, and metabolic 19 Mackenzie EJ, Rivara FP, Jurkovich GJ, et al. The national study on costs and outcomes risks or clusters of risks for 195 countries and territories, 1990–2017: a systematic of trauma. J Trauma 2007;63:S54–67. analysis for the global burden of disease study 2017. The Lancet 2018;392:1923–94. 20 Traumatic Brain Injury(TBI). Follow-­Up registry and surveillance of TBI in the 6 Roth GA, Abate D, Abate KH, et al. Global, regional, and national age-sex-­ ­specific emergency department (ED); notice of availability of funds. federal register, 2002. mortality for 282 causes of death in 195 countries and territories, 1980–2017: Available: https://www.​federalregister.gov/​ ​documents/2002/​ ​05/08/​ ​02-11359/​ ​ a systematic analysis for the global burden of disease study 2017. The Lancet traumatic-brain-​ ​injurytbi-follow-​ ​up-registry-​ ​and-surveillance-​ ​of-tbi-​ ​in-the-​ emergency-​ ​ 2018;392:1736–88. department-​ed [Accessed 14 May 2018]. 7 Dicker D, Nguyen G, Abate D, et al. Global, regional, and national age-­sex-­specific 21 Fullman N, Yearwood J, Abay SM, et al. Measuring performance on the healthcare mortality and life expectancy, 1950–2017: a systematic analysis for the global burden access and quality index for 195 countries and territories and selected subnational of disease study 2017. The Lancet 2018;392:1684–735. locations: a systematic analysis from the global burden of disease study 2016. The 8 Foreman KJ, Naghavi M, Ezzati M. Improving the usefulness of US mortality data: new Lancet 2018;391:2236–71. methods for reclassification of underlying cause of death. Popul Health Metr 2016;14. 22 Salomon JA, Haagsma JA, Davis A, et al. Disability weights for the global burden of 9 Naghavi M, Makela S, Foreman K, et al. Algorithms for enhancing public health utility disease 2013 study. Lancet Glob Health 2015;3:e712–23. of national causes-­of-death­ data. Popul Health Metr 2010;8:9. 23 Lozano R, Fullman N, Abate D, et al. Measuring progress from 1990 to 2017 and 10 GBD 2016 Traumatic Brain Injury and Spinal Cord Injury Collaborators. Global, regional, and projecting attainment to 2030 of the health-­related sustainable development goals national burden of traumatic brain injury and spinal cord injury, 1990-2016: a systematic for 195 countries and territories: a systematic analysis for the global burden of analysis for the global burden of disease study 2016. Lancet Neurol 2019;18:56–87. disease study 2017. The Lancet 2018;392:2091–138. 11 SpringerLink. Intentional injuries in the eastern Mediterranean region, 1990–2015: 24 Sustainable Development Goals. Sustainable development knowledge platform. findings from the global burden of disease 2015 study, 2015. Available: https://​link.​ Available: https://​sustaina​ ​bledevel​ ​opment.un.​ ​org/?menu=​ ​1300 springer.​com/​article/ [Accessed 29 Aug 2019]. [Accessed 27 Jul 2019]. http://injuryprevention.bmj.com/ on September 27, 2021 by guest. Protected copyright.

James SL, et al. Inj Prev 2020;26:i125–i153. doi:10.1136/injuryprev-2019-043531 i153